An Economic Analysis of The

The Center for Law and Economic Studies
Columbia University School of Law

435 West 116th Street
New York, NY 10027-7201
(212) 854-3739
An Economic Analysis of the Guaranty Contract
Avery W. Katz
Professor of Law
Working Paper No. 136
A revised version of this working paper has been published in

University of Chicago Law Review, Vol. 66, No. 1, Winter 1999
Please send comments to:
Avery W. Katz
Georgetown University Law Center
600 New Jersey Avenue, NW
Washington, DC 20001
tel: (202) 662-9375
E-mail: katz@law.georgetown.edu
Do not quote or cite without author's permission.
This paper can be downloaded without charge at:
The Columbia Law School Working Paper Series Index:

http://www.law.columbia.edu/lawec
The Social Science Research Network Electronic Paper Collection:

http://papers.ssrn.com/paper.taf?abstract_id=43104
Avery Wiener Katz

Georgetown University
Draft version: March 1998
Abstract: Guaranty arrangements, in which one person stands as surety for a second person’s
obligation to a third, are ubiquitous in commercial transactions and in commercial law. In recent years,
however, scholarly attention to the topic has been scant; and there is still no theoretical treatment of this
body of law or practice from a economic policy perspective. This paper, accordingly, attempts to
outline the basic economic logic underlying the guaranty relationship, and applies the results to a
variety of specific issues in government policy and private planning. It poses and answers three main
questions: First, why would a creditor prefer to make a guaranteed loan rather than an unguaranteed
one? The answer is not as obvious as might first appear, given that market competition over credit
terms tends to adjust the interest rate paid by an individual borrower to reflect the specific default risk
that he presents. Second, given that they bear the residual risk of debtor default, why would guarantors
prefer to guarantee loans rather than make loans directly, thus foregoing the opportunity to earn interest
payments that could help to compensate for the risk they bear? Third, even if it is efficient for one
creditor to provide funds and another to provide insurance against default, why would the parties prefer
to implement this arrangement through the triangular form of a guaranty, instead of simply having the
former creditor lend to the latter and the latter lend to the ultimate borrower?
Keywords: Contracts, commercial law, guaranties and suretyships, risk allocation.
Address: Until 5/31/98: Columbia University School of Law, 435 W. 116th St., New York, NY
10027. Phone: (212) 854-2599. FAX: (212) 854-7946. After 6/1/98: Georgetown
University Law Center, 600 New Jersey Avenue NW, Washington, DC 20001.
Phone: (202) 662-9375. FAX: (202) 662-9410.
E-mail: katz@law.georgetown.edu
Avery Wiener Katz
I. Introduction
Guaranty arrangements, in which one person stands as surety for a second person’s obligation to
a third, are ubiquitous in commercial transactions and in commercial law. They are at least as
common, if not more so, as personal property security or real mortgages. The interrelated rights and
duties arising out of guaranties and analogous legal and financial devices comprise a substantial part
of several distinct bodies of law, including those governing secured transactions, negotiable
instruments, letters of credit, and contractual assignments; and the distinctions among guaranties and
alternate transactional forms are relevant to a wide array of regulatory regimes, including tax,
banking, insurance, and securities law. Up until the 1950's, indeed, the standard law school
curriculum even included a required course on the subject, under the title of suretyships.
In more recent years, however, the law of guaranties has languished in a scholarly backwater.
Lawyers specializing in the area lament that the field has “fallen between the cracks,”1 and that bench,
bar, and academy are all ignorant of its content and practical importance.2 From a purely doctrinal
viewpoint, fortunately, this situation is beginning to change. Three major new reformulations of
guaranty law — the recent revisions of Articles 3 and 5 of the Uniform Commercial Code and the
just–released Third Restatement of Suretyship and Guaranty — have been adopted in the last several
years. Practicing lawyers will be forced to familiarize or re–familiarize themselves with the legal rules
in this area, at the risk of liability for malpractice. But despite this flourishing of doctrinal work, and
Professor of Law, Georgetown University Law Center, and Visiting Professor of Law, Columbia University School
of Law. I am grateful to Allan Axelrod, Robert Barsky, Lisa Bernstein, Jim Bowers, Merritt Fox, Michael Heller,
Claire Hill, Wendy Wiener Katz, Michael Klausner, Kyle Logue, Ronald Mann, Scott Masten, Warren Schwartz,
William Vukowich, James J. White, and seminar participants at the University of Michigan, Columbia, and
University of Pennsylvania law schools and at the American Law and Economics Association for helpful
comments, to Andrew Niebler and John Sheckler for excellent research assistance, and to the Dean’s Research
Fund at Georgetown University Law Center for financial support. This paper is a preliminary draft and is subject
to revision.
1. Donald J. Rapson, History and Background of the Restatement of Suretyship, 34 WM. & MARY L. REV. 989,
1011 (1993).
2. See, e.g., Rapson, supra note 1; Neil B. Cohen, Striking the Balance: The Evolving Nature of Suretyship
Defenses, 34 WM. & MARY L. REV. 1025, 1025-26 (1993); Gerald T. McLaughlin, Standby Letters of Credit
and Guaranties: An Exercise in Cartography, 34 WM. & MARY L. REV. 1139, 1139 (1993).
Katz, Economics of the Guaranty Contract Draft, 3/98, p. 2
despite the commercial importance of the transactions that the field affects, there has still been no
systematic theoretical treatment of the subject from a economic policy perspective. This omission
is especially noteworthy, moreover, given the substantial body of policy and theoretical scholarship
in recent years in the cognate fields of secured transactions and bankruptcy. While economists and
economically influenced lawyers have written on such topics as the economic function of secured
debt,3 of bankruptcy law’s function in preserving creditors and debtors with incentives to maximize
the value of failing enterprises,4 and of the incentive and notice problems created when creditors take
nonpossessory interests in a debtor’s property,5 there has been no general analysis of the simple yet
widespread guaranty contract, for which all these other devices are economic substitutes.
The purpose of this paper is to outline the basic economic logic of the guaranty relationship —
its underlying costs and benefits, and the key features that distinguish it from alternative financial
arrangements. This basic logic is simple, but it has application across a broad variety of problems
faced by commercial and business lawyers. Among these problems are: giving content to and
determining the scope of suretyship rights and defenses across diverse doctrinal fields, reconciling
differences and inconsistencies among suretyship rules in adjoining fields of law; and helping to
identify the functional essence of suretyship for purposes of independent regimes of regulation such
as tax and banking, so that those regimes are better able to treat like cases alike according to their
own individual policy objectives. Most importantly, however, a clear account of the economics of
guaranties is essential to understanding why contracting parties might, and when they should, enter
into guaranty relations or into their individual incidents. Such an understanding is critical to an
designing an intelligent regulatory regime for debtor–creditor relations, but it is even more so for the
lawyer engaged in advising clients how structure their financial affairs. In most situations, clients
3. See, e.g., Alan Schwartz, Security Interests and Bankruptcy Priorities: A Review of Current Theories, 10 J.
LEGAL STUD. 1 (1981); Robert E. Scott, A Relational Theory of Secured Financing, 86 COLUM. L. REV. 901
(1986); Alan Schwartz, A Theory of Loan Priorities, 18 J. LEGAL STUD. 209 (1989); George G. Triantis,
Secured Debt Under Conditions of Imperfect Information, 21 J. LEGAL STUD. 225 (1992).
4. See, e.g., Thomas Jackson, THE LOGIC AND LIMITS OF BANKRUPTCY LAW (1986); Frank H. Buckley, The
Bankruptcy Priority Puzzle, 72 VA. L. REV. 1393 (1986); Douglas G. Baird, Revisiting Auctions in Chapter
11, 36 J. LAW & ECON. 633 (1993); Douglas G. Baird and Randal C. Picker, A Simple Noncooperative
Bargaining Model of Corporate Reorganizations, 20 J. LEGAL STUD. 311 (1991).
5. See, e.g., Douglas G. Baird, Notice Filing and the Problem of Ostensible Ownership, 12 J. LEGAL STUD. 53
(1983); ; Douglas G. Baird and Thomas H. Jackson, Possession and Ownership: An Examination of the Scope
of Article 9, 35 STAN. L. REV. 175 (1983); Saul Levmore, Variety and Uniformity in the Treatment of the
Good-faith Purchaser, 16 J. LEGAL STUD. 43 (1987).
have some leeway to choose whether to pursue their goals through guaranties or through alternative
devices; in those cases in which they do decide to enter into guaranties, they typically have freedom
to contract out of the various default rules with which suretyship law provides them.6 A good
commercial lawyer thus needs to understand the functional underpinnings of the transaction in order
to help plan it — and in commercial settings, these underpinnings are economic.
Such a functional perspective is relevant in both the private and public spheres. For example,
many government policies are intended to assist debtors, both individual and institutional. These
policies, which include federal student loans, government–backed mortgages or farm loans, bailouts
of private entities such as Chrysler or Lockheed, and international loan guaranties for foreign allies
such as Israel or Mexico, are motivated by efficiency, distributional, or other considerations.
Whatever the goals of such programs, however, it is in the public interest that they be pursued
through an efficient transactional form. For instance, in the last few years federal support for
education lending has shifted from a program of guaranteed student loans to one of direct student
loans, on the supposed rationale of greater administrative efficiency; and a third alternative — the
Clinton administration’s various proposals for new tax subsidies for student college loans — played
a minor role in the 1996 presidential campaign.7 An economic perspective on guaranties can help
shed light on which of these programs would be cheapest or most effective.
Nonetheless, the arguments to follow may well be of greater utility to business lawyers designing
individual transactions than to legislators, judges, or public regulators. In my view, however, this is
a strength of the analysis rather than a weakness. As I have argued elsewhere, contemporary legal
scholars, even those associated with the relatively market–oriented perspective of law and economics,
have focused disproportionately on analyzing the merits of particular public policies and have
6. See e.g., UNIFORM COMMERCIAL CODE 1–102(3) [providing that rules of UCC may in general be varied by the
parties’ agreement, and that obligations of reasonableness, diligence, and good faith, while not disclaimable,
can be given specific content through agreement.]; RESTATEMENT (THIRD) OF SURETYSHIP AND GUARANTY
§6 (1996) [both substantive and procedural rules of Restatement variable by agreement].
7. For surveys and critiques of recent changes and proposed changes in federal policy toward student loans, see
generally Evelyn Brody, Paying Back Your Country Through Income-Contingent Student Loans, 31 SAN
DIEGO L. REV. 449 (1994), Barbara Miles and Dennis Zimmerman, Substituting Direct Government Lending
for Guaranteed Student Loans: How Budget Rules Distorted Economic Decisionmaking, 47 NAT. TAX J. 773
(1994); Dennis Zimmerman and Barbara Miles , Reducing Costs and Improving Efficiency in the Student
Loan Program 50 NAT. TAX J. 541 (1997).
addressed themselves primarily to government decisionmakers.8 Because private actors also care
about efficiency and distribution in ordering their affairs, however, the theoretical insights of
economics are at least as valuable to them as they are to public officials. Just as the citizens of the
United States must decide whether, through their federal government, to lend funds to the
government of Mexico directly, to guarantee loans made to Mexico by a third-party lender, or to
leave Mexico to its own financial devices, so must a commercial bank decide whether to demand that
a credit applicant provide a co–signer, or whether to issue a standby letter of credit for one of its
existing customers who is seeking additional funds from another lender. The same set of economic
and functional issues are relevant whether one is making public or private policy.
In the succeeding sections of this paper, therefore, I lay out the basic economics of suretyships
and guaranties. The paper poses three main questions. First, why would a creditor prefer to make
a guaranteed loan rather than an unguaranteed one? The answer is not as obvious as might first
appear, given that market competition over credit terms tends to adjust the interest rate paid by an
individual borrower to the specific default risk that he or she presents, Second, why, given that they
bear the residual risk of debtor default, would guarantors prefer to guarantee loans rather than make
loans themselves, thus foregoing the opportunity to earn interest payments that could help to
compensate for this risk? Third, even if it is efficient for one lender to provide funds and another to
provide insurance against the borrower’s default, why would the parties prefer to accomplish such
an arrangement through the three–corner form of a guaranty, instead of simply having the former lend
to the latter and the latter on–lend to the ultimate borrower?
In brief, the answers to these questions are as follows. First, and most straightforwardly, creditors
should prefer to make guaranteed loans when, and only when, there is a potential guarantor who can
monitor the debtor more cheaply than the creditor can. Monitoring in this regard should be
understood broadly; as I use the term it is intended to incorporate any and all activities that reduce
the expected cost of nonpayment, including up–front investigation and underwriting, assessing the
merits of the project the debtor proposes to finance, supervising the debtor and his project during the
life of the loan, and enforcement, collection, and salvage in the event of default. Second, the person
who is the lowest–cost monitor will prefer to act as guarantor, rather than as primary lender, when
and only when her cost of liquidity — that is, the transaction costs she must pay to convert her assets
into a acceptable means of ready payment — is less than that of potential third-party lenders. In
8. See Avery Katz, Taking Private Ordering Seriously, 144 U. PENN. L. REV. 1745 (1996).
essence, suretyship is an economic arrangement in which one party specializes in providing liquid
funds, and another specializes in providing informational monitoring and insurance. Third, and
perhaps least obviously, guaranties are preferable to on–lending or other forms of intermediation
when, and only when, the party who provides liquid funds attaches a higher value to the option to
proceed directly against the ultimate borrower than does the party who provides monitoring — or,
alternatively, than do the monitor’s other creditors.
To develop these ideas more fully, however, requires some groundwork. The succeeding section
of this paper, accordingly, presents some background on the subject of guaranties. It sets out a
variety of stereotypical situations giving rise to guarantor liability, outlines the various bodies of law
that govern such situations, and summarizes the main questions and unresolved controversies that
arise in those legal fields. Section III, which forms the heart of the paper, then compares the
economic costs and benefits of guarantees with those of related transactional forms, and discusses
why contracting parties would — and should — prefer one type of transactional form to another.
Section IV then applies the theoretical analysis to two questions of public policy and private planning:
how should the government operate when it wishes to subsidize borrowing, and what risks and duties
should be encompassed within the guaranty that is made by a commercial bank upon issuing a standby
letter of credit. Section V summarizes the analysis, suggests some further applications to analogous
problems in law, policy, and planning, and offers general conclusions.
II. The law of guaranties and suretyships: a sense of the terrain
The number and variety of transactions that include or imply guaranty relationships are vast. In
order to situate the problem in a business and policy context, this section surveys a sample of such
relationships and summarizes the legal issues that arise from them.
A. The stereotypical guaranty relationship
Guaranties and suretyships take many forms, but the essence of the relationship is a contract
among three parties: a creditor C, who is owed a duty, a debtor D, who owes the duty, and a
guarantor or surety G, who promises to perform or pay damages on D’s behalf.9 Depending on the
9. Traditionally, a “suretyship” describes a relationship in which the guarantor and debtor are jointly and severally
liable on the underlying obligation, while under a “guaranty,” the debtor’s actual default is a condition
Figure 1: Terminology of the guaranty relationship in alternative contexts
C D G
Restatement of Suretyship Obligee Primary obligor Secondary obligor
Negotiable instrument Holder Accommodated party Accommodation party
Sale of accounts Assignee Account debtor Assignor
Letter of credit Beneficiary Applicant Issuer
Lease assumption Landlord Subtenant Primary tenant
context and the applicable body of law, however, these three parties are designated by different labels,
as illustrated in Figure 1. In the terminology of the new Restatement (Third) of Suretyship and
Guaranty, for instance, C is deemed the “obligee,” D the “primary obligor,” and G the “secondary
obligor.”10 As this terminology suggests, as between D and G, D is expected to bear the burden of
performance, and if G is called upon to perform or pay in D’s stead, it is ordinarily expected that D
will reimburse her if he is able.11 The arrangement can be achieved through a single contract or
through multiple contracts; in the latter case, G may contract with D alone, with C alone, or with
both. It is even possible for a guaranty to be created without the knowledge of D or C, though in this
case certain of the parties’ rights and duties will be altered in order to protect the uninformed party.
precedent to the guarantor’s obligation to perform. See RESTATEMENT (THIRD) OF SURETYSHIP AND
GUARANTY, §15 (1996). Due to variations among provisions of individual guaranty or suretyship contracts,
however, the two terms overlap substantially in actual practice. Id., comment b; see also id., §1, comment c.
For this reason, I will use the terms interchangeably in this paper, except where the question of a condition
precedent to guarantor liability is specifically relevant to the discussion.
10. RESTATEMENT (THIRD) OF SURETYSHIP AND GUARANTY, §1 (1996). The new Restatement is intended to
supercede its most immediate predecessor, Division II of the Restatement of Security (1941). It also covers
certain topics treated in the Restatement (Second) of Contracts.
11. In order to avoid ambiguity, when discussing generic guaranty transactions in this paper I will adopt the
convention of referring to debtors and primary obligors with male pronouns, guarantors and sureties with
female pronouns, and creditors with neuter pronouns.
This formal relationship is general enough to encompass a wide variety of business and policy
contexts, as the following examples demonstrate.
1. Interfamilial guarantees
D, a college senior, is about to graduate and take his first job, and wants to buy a car. He has
enough money for the down payment, but has not yet established a credit history. His only
experience with borrowing consists of the student loans he took out to finance his education, which
will not come due for six months and which, accordingly, he has not yet begun to repay. Because of
this situation, the only finance companies that will deal with him on his own are those that specialize
in lending to risky debtors, and those companies’ interest rates are substantially higher than the
market average. If he can get his loan guaranteed by someone with an established credit record, on
the other hand, he will be eligible for more conventional financing from C, a local bank. D arranges
with his mother, G, to provide such a guaranty by co–signing his loan application. At C’s request,
G also co–signs the promissory note that D executes in C’s favor before receiving funds.
2. Intercorporate guaranties
D, a limited liability corporation engaged in the production of baked goods, needs financing to
modernize its capital equipment. C, a producer of cereal food products and D’s longstanding supplier
of flour, is willing to lend D the necessary funds, but is concerned about the value of D’s marketable
assets and about having to share them with other creditors in the event that D runs into financial
trouble. As a condition of the loan, accordingly, C requests and obtains guaranties from D’s parent
company, G1, and from G1's sole shareholder, G2. In the commercial setting, furthermore, it is also
common for such guaranties to be accompanied by other contractual covenants and protections for
the lender. For example, both the underlying obligation and the guaranties might be secured by a
purchase money security interest in the new machinery, or by mortgages on real property owned by
D, G1, or G2.12
3. Standby letter of credit
D, a software design company, wishes to borrow funds for a new project from C, a commercial
bank with whom it has not previously dealt. Because this would entail a new credit investigation and
12. The example is taken from the contract that formed the basis of the dispute in In re Cushman Bakery, 526 F.2d
23 (1st Cir. 1975).
C’s familiarizing itself with D’s business, C would have to charge D more than the prime rate on the
loan. Instead, D goes to G, another bank with whom it has a longstanding relationship, and applies
for a standby letter of credit naming C as beneficiary. This standby letter provides that if C presents
G with certain documents attesting that D has defaulted on the loan, C will be entitled to draw on the
letter to the extent of any funds it has at risk. In exchange for a payment in an amount proportional
to the funds being lent, G issues the standby letter to D, who presents it to C. On the strength of the
letter, C lends D the funds for the project at or below the prime rate.13
4. Construction suretyships
D, a construction contractor, submits the lowest bid on a job to build a new school on a site
owned by C, a municipal school district. As a condition of awarding D the general contract on the
school, C requires that D obtain a performance bond from a third–party surety insurer. D purchases
such a bond from G, a company specializing in providing such insurance. Under the terms of the
bond, G promises to undertake to complete the construction job if D does not, to pay all outstanding
amounts owed to subcontractors and suppliers of materials, to remove any construction or
mechanics’ liens on the property at the conclusion of the project, and to pay any penal sums owing
to C under the underlying construction contract. By its terms, the performance bond also provides
that all funds owed by C to D under the construction contract stand as collateral for any claims that
G may acquire against D. Thus, if G completes the job in D’s stead, G, not D, will be entitled to
receive the contract price.14
5. Statutory and fidelity bonds
Under various state and federal statutes, persons engaged in activities that are deemed especially
likely to result in tort or criminal liability, or who take on substantial fiduciary duties, are required to
post a bond sufficient to cover possible damages and monetary punishment. Bail bonds are a classic
13. See generally McLaughlin, supra note 2; Bernard S. Wheble, "Problem Children" — Standby Letters of Credit
and Simple First Demand Guarantees, 24 ARIZONA L. REV. 301 (1982); Dean Pawlowic, Standby Letters of
Credit: Review and Update, 23 U.C.C. L. J. 391 (1991); Henry Harfield, Guaranties, Standby Letters of
Credit, and Ugly Ducklings, 26 U.C.C. L. J. 195 (1994).
14. Performance bonds such as these are a typical provision in government contracts at the state and local level,
and are required for all federal public works projects under the Miller Act, 40 U.S.C.S. §§270a et seq. (1997).
See generally T. Scott Leo, The Construction Contract Surety and Some Suretyship Defenses, 34 WM. &
MARY L. REV. 1225 (1993).
example, but similar “fidelity bonds” are commonly required to be posted by a wide assortment of
public officers and employees, by persons engaged in potentially hazardous activities such as toxic
waste disposal, by ordinary tradepersons such as plumbers and roofers, and by officers and employees
of financial institutions. While such persons may not be debtors in the immediate sense, they are
contingently so, and need to set aside funds for the contingency. Instead of posting such bonds out
of their own funds, however, many such contingent debtors arrange for bonding by a third–party
guarantor in exchange for an up–front payment or insurance premium. In this case, the creditor C
is the person or persons who would be entitled to payment should the debtor actually incur liability.15
6. Recourse assignment of accounts, chattel paper, or instruments
G, a dealer in consumer appliances, sells a refrigerator to D on credit. G then assigns her interest
in D’s account, together with a number of other outstanding accounts, to C, a factor (i.e., a company
in the business of buying such accounts), in exchange for an immediate cash payment. The underlying
consumer debt may also be secured by an security interest in the refrigerator, in which case the
assigned contract is referred to as “chattel paper;” alternatively or additionally, it may be embodied
in the form of a negotiable promissory note signed by D as maker and indorsed by G over to C. In
any of these cases, however, the amount of cash that C pays G for the account, chattel paper, or
instrument will be less than its face value. This discount reflects the facts that payment will not be
forthcoming for a time and that the factor must service the account in the interim and incur costs of
collection. The assignment, however, is on a recourse basis: to the extent that the accounts are
uncollectible (or more generally, to the extent that the shortfall in collections exceeds some agreed
amount), G must refund C’s payment and retake possession of the bad account. This effectively
renders G a guarantor of the underlying debt. A similar analysis applies to assignments of contractual
rights more generally, so long as the assignment is with recourse.16
15. See generally Hugh E. Reynolds, Jr. and James Dimos, Fidelity Bonds and the Restatement, 34 WM. & MARY
L. REV. 1249 (1993); James A. Black, Jr., Miscellaneous Surety Bonds and the Restatement, 34 WM. & MARY
L. REV. 1195 (1993).
16. See generally D. Benjamin Beard, Suretyship on the Fringe: Suretyship by Operation of Law and by Analogy,
34 WM. & MARY L. REV. 1195 (1993).
7. Assumption of leases and real estate mortgages
G holds a thirty–year lease on a downtown retail location owned by C at a rent substantially more
favorable than current market rates. G wishes to move to a different location in order to expand her
business, but she is reluctant to surrender the valuable lease. Instead, she sublets the premises to D,
who agrees to assume her obligations to C under the original lease, and in addition to pay an
additional monthly rent to G. C consents to the sublease in exchange for a side payment, but does
not release G from the terms of the original lease. Thus, if D fails to pay the primary rent, C is
entitled to collect it from G. A similar analysis applies to delegations of contractual duties more
generally.17
8. Joint and several liability
Not only might there be multiple guarantors and multiple debtors in a single transaction, in some
transactions parties might be debtors as to part of an obligation and guarantors as to the rest.18
Consider the following example: two law students, A and B, decide to room together and lease an
apartment. The total monthly rent is $1000, but the students agree between themselves that A will
take the larger bedroom and pay $600/month, and B will take the smaller bedroom and pay
$400/month. The lease contract, however, provides that each tenant is jointly and severally liable for
the entire rent. This provision renders A and B both primary obligors and guarantors with respect
to their landlord. As to the $600 rent on the larger bedroom, A is primary obligor and B is guarantor;
as to the $400 rent on the larger bedroom, B is primary obligor and A is guarantor. This is so
whether or not the landlord knows of the students’ side agreement; indeed, one advantage of such
an arrangement is that the landlord does not need to keep track of how its tenants apportion the lease
obligations between themselves.
9. Government–sponsored enterprises
A number of U.S. government–chartered and sponsored institutions participate directly in credit

markets. Among these are the Federal National Mortgage Association (“Fannie Mae”), the Federal
Home Loan Mortgage Corporation (“Freddie Mac”), the Farm Credit System, and the Student Loan
17. Id. at __.
18. See RESTATEMENT (THIRD) OF SURETYSHIP AND GUARANTY, §§ 52–70 (1996) [“Multiple Secondary
Obligors”].
Marketing Association (“Sallie Mae”). In some instances these institutions operate as direct lenders,
but in many other instances (especially when they issue mortgage–backed or student–loan–backed
securities) they operate as guarantors of loans extended by private lenders. As indicated above, there
has been substantial policy discussion in recent years over what mix of loans, guaranties, and grants
would best serve home buyers, farmers, and students who are the ultimate beneficiaries of these
programs.
Furthermore, while these various entities are now technically nongovernmental agencies, as a
historical matter all were created and periodically infused with federal funding, and they continue to
be subject to substantial federal regulation.19 While they have no formal legal claim on the assets or
the general revenues of the federal government, their importance to the national economy — and
more significantly, their political popularity — is such that if any one of them were in danger of
failure, Congress would come under substantial pressure to step in and rescue it from default. In an
informal political sense, accordingly, U.S. taxpayers stand as the ultimate guarantors of this debt.20
For this reason, the investor community is willing to lend to these entities at lower interest rates than
it otherwise would demand from a comparable private borrower, notwithstanding the informality of
this guaranty.
10. Guaranties between independent national sovereigns
On several recent occasions, the U.S. government has assisted allied countries in obtaining private
sector financing for their national debts, both as a form of foreign aid and as a tool for pursuing other
foreign policy objectives. For instance, in the fall of 1992 the U.S. guaranteed $10 billion in new
loans taken out by the government of Israel for the purpose of financing housing for emigrants from
the former Soviet Union; and in January 1995 it guaranteed $40 billion in loans to Mexico as part of
an international aid package designed to prevent the Mexican government from defaulting on existing
obligations after market pressures forced it finally to devalue the peso. In both instances the loan
guaranties were the subject of substantial public controversy, both on fiscal and on foreign policy
19. See generally the REPORT OF THE SECRETARY OF THE TREASURY ON GOVERNMENT–SPONSORED
ENTERPRISES (1991).
20. For a discussion of the potential financial problems raised by the federal government’s responsibility for these
entities, see id. at __.
grounds.21 Much of the controversy had to do with whether the U.S. should be extending aid to these
beneficiary countries at all, as well as with which domestic actors should have the authority to make
the decision, but some had to do with the guaranty form. Similarly, the U.S. government has
routinely guaranteed smaller loans for specific purchases such as arms sales, both for foreign policy
reasons and in order to assist American commercial interests in those areas.22
B. The law governing guaranties
Because guarantor liability can be achieved through a variety of contractual devices and
commercial specialties, the precise body of law that governs will depend on which device is used.
Most generally, the parties’ relationship is governed by the common law. In the guaranty context,
this includes not just the common law of contracts, but also the special and less widely known rules
of the common law of suretyship; these latter doctrines have recently been recatalogued and
republished in the new Restatement (Third) of Suretyship and Guaranty. If a guarantor G acquires
her status by making or indorsing a negotiable instrument, however, then the parties’ relationship is
governed first by Article 3 of the Uniform Commercial Code, which provides special rules for
so–called “accommodation parties,” and next by any common-law rules that the UCC does not
specifically displace.23 If the guaranty takes the form of a standby letter of credit, as is routine in
transactions in which the guarantor is a commercial bank, the relationship is governed in part by
Article 5 of the UCC, in part by the common law, and in most transactions with an international
element, by the International Chamber of Commerce’s Uniform Customs and Practice for
21. See, e.g., John Quigley, Loan Guarantees, Israeli Settlements, and Middle East Peace, 25 VAND. J.
TRANSNATIONAL L. 547 (1992) (arguing that such loan guaranties, as well as other forms of US aid to Israel
facilitate and may themselves constitute violations of international law, and diminish prospects of peace in
Mideast); James D. Humphrey, II, Student Note: Foreign Affairs Powers and "The First Crisis of the 21st
Century": Congressional vs Executive Authority and the Stabilization Plan for Mexico, 17 MICH. J. INT'L L.
181 (1995) (describing and analyzing political contest between President and Congress over Mexican
guaranties).
22. See generally Jonathan E. Sanford, Foreign Debts to the U.S. Government: Recent Rescheduling and
Forgiveness 28 G.W. J. INT'L L. & ECON. 345 (1995)
23. See UCC §3–419 (1995) [”Instruments Signed for Accommodation”]; UCC §3–605 (1995) [”Discharge of
Indorsers and Accommodation Parties.“] These sections apply only if an indication of G’s guaranty appears
on the face of the instrument being guaranteed. In this case, G is referred to as an “accommodation party,”
and D as an “accommodated party.” “Off–the–paper” guaranties, conversely, are governed solely by the
common law. See generally Neil B. Cohen, Suretyship Principles in the New Article 3: Clarifications and
Substantive Changes, 42 ALA. L. REV. 595 (1991).
Documentary Credits.24 If the guarantor provides collateral to secure either the primary or the
secondary obligation, and in all cases involving the sale or assignment of accounts, then Article 9 of
the UCC provides various rules governing the parties’ respective rights and duties. The relationship
could also be governed by state and federal statutes regulating the sale of financial options, because
one common way to provide the functional equivalent of a guaranty is for G to sell C a put option
on the right to collect from D, which C will subsequently wish to exercise if and only if the expected
value of D’s performance falls below the exercise price. If the debtor files for federal bankruptcy
protection, a variety of Bankruptcy Code provisions (the most important of which deal with
preferences and fraudulent conveyances) also apply to the transaction; and outside of bankruptcy state
rules of debtor-creditor law, including the law of fraudulent conveyances, also govern. And in the
case of loans guaranteed by the US and by U.S.–sponsored institutions, either to private individuals
or to other sovereign governments, federal statutes — and ultimately, implicit political contracts
among Congress, the executive branch, international actors, and American voters — provide the basis
for liability.
Whatever governing body of law, practice, or political understanding guaranties flow from,
however, the functional issues they raise are much the same. Guaranties are a response to potential
moral hazard and adverse selection on the part of the debtor; they help protect creditors against some
of the risks of debtor misbehavior or insolvency by shifting those risks to guarantors. In so doing,
they also enlist the guarantor’s efforts in reducing or managing those risks. As with any other
insurance arrangement, however, guaranties are themselves subject to various types of moral hazard
and adverse selection; each of the parties to the transaction can impose losses on the others by failing
to take precaution or by withholding information. For example, guarantors have the power to hurt
creditors (and perhaps debtors as well) by incautiously offering guaranties where the underlying
financial situation does not warrant it, or by failing to monitor the debtor after money has been lent.
Creditors may hurt guarantors (and perhaps debtors) by provoking debtors to default, by failing to
collect from the debtor where possible, and by impairing the guarantor’s ability to seek recourse from
the debtor after the fact. Debtors can injure guarantors and creditors alike by taking excessive risks
and making inadequate efforts to meet their primary obligations. Moreover, the very creation of a
24. For a survey of international law and practice in the area of letters of credit, see Boris Kozolchyk, Bank
Guarantees and Letters of Credit: Time for a Return to the Fold, 11 U. PA. J. INT’L BUS. L. 1 (1989). The
extent to which common–law principles should continue to govern letters of credit is a matter of dispute, as
section IV.B, infra, discussses. See generally Peter A. Alces, An Essay on Independence, Interdependence,
and the Suretyship Principle, 1993 U. ILL. L. REV. 447 (1993) (arguing that treating letters of credit separately
from ordinary suretyships is both a doctrinal and conceptual mistake).
suretyship can itself raise incentive problems, particularly when one of the parties to the underlying
primary contract had not anticipated that a three–corner relationship would arise. For instance, a
creditor who privately and unilaterally arranges for a guaranty may substantially increase the debtor’s
cost of compliance under the primary contract, multiplying the number of persons to whom he owes
duties of care or disclosure. Similar problems arise when a creditor transfers its interest in the
debtor’s account to a distant and unfamiliar assignee, or when a debtor delegates performance of
contractual duties to an untested person with whom the creditor will nonetheless be obligated to make
reasonable efforts to deal.
Because of these common functional problems, therefore, all bodies of formal guaranty law
provide the three archetypical parties with various rights and duties that, on the whole, are designed
to discourage opportunism and to encourage all persons concerned to take reasonable precautions
against economic loss. Thus, all regimes of guaranty law provide as a default rule that if the debtor
does not perform his primary obligation and the guarantor has to make good on her guaranty, then
she will have a right of recourse against him. Under the common-law of suretyship, for instance, this
right of recourse is protected in three ways; first, the guarantor is entitled to a right of reimbursement
for any funds she actually pays out, plus reasonable expenses.25 Second and alternatively, if G
satisfies D’s obligation to C in full, she is equitably subrogated to C’s rights against him.26 This
subrogation right is valuable if C has some special status (such as that of holder in due course) that
confers immunity to claims and defenses that D could otherwise raise (and specifically, that D could
raise against G); it is also useful if C has priority over D’s other creditors through a lien or other
interest in specific property. Third, G has a right to D’s performance, or what has sometimes been
called the right of exoneration.27 This right is valuable because G can assert it even before it has been
called upon to pay anything out, and such early action may increase the chances that D can still
perform. Depending on the circumstances, exoneration may entitle G to specific performance from
D, to an action against D for creating insecurity along the lines of the right of adequate assurances
under §2–609 of the UCC, or to a right to declare the contract in default.
25. See Restatement (Third) of Suretyship and Guaranty, §22 (1996); cf. UCC §3-419(e) (1995); UCC §5-114
(1995).
26. See Restatement (Third) of Suretyship and Guaranty, §29 (1996).
27. See id. §21.

Conversely, guaranty law provides a number of rules designed to protect guarantors against
creditor action that materially increases guarantor risk. Thus, if the debtor’s breach of duty was
occasioned by a countervailing breach by C — as when a credit buyer of goods withholds payment
after discovering that the goods are unmerchantable — G is entitled under traditional common-law
rules to set up C’s breach as a defense to her secondary liability to the same extent it would provide
a defense to D’s primary liability.28 Similarly, she may be able to assert certain unrelated claims
belonging to D by way of a set–off.29 Beyond such derivative claims, furthermore, G is entitled to
the benefit of certain rights and defenses against C that D could not himself raise; these are the
so–called “suretyship defenses.” For instance, if C modifies D’s underlying obligation, extends D’s
time for performance, unreasonably refuses a tender of performance by either D or G, fails to
supervise or preserve rights against collateral or against co–sureties and sub–sureties, G may be
discharged from guarantor liability to the extent that she suffers resulting loss.30 Indeed, if C fails to
disclose events that would support a good suretyship defense, under conditions where it should know
of G’s ignorance of such events, G has a right to recover any amounts she has mistakenly and
unnecessarily paid out.31
The specific content of these various rights and duties, however, depends both upon the particular
factual setting and upon the governing body of law. The substance of suretyship defenses and the
duties of care owed by the creditor, for instance, are different under the common law of suretyship,
Articles 3 and 9 of the UCC, and the law of letters of credit.32 So are the remedies that follow from
28. See id. §34.
29. See id. §36.
30. See id. §37 et seq. (“Impairment of Suretyship Status”).
31. Id. §47 (1996).
32. For instance, the creditor’s release of the debtor presumptively releases the guarantor under the traditional
common law of suretyship, but does not operate as a release under Article 3; compare Restatement (Third) of
Suretyship and Guaranty §39 (1996) with UCC 3–605(b). Similarly, under the common law, the guarantor
can raise most defenses against liability that the debtor would be able to, including failure of consideration,
breach of warranty, and the like. Under Article 3, however, such defenses would not be good as against a
holder in due course, and under new Article 5, they would be good only to the extent that they amounted to
material fraud under 5–109. See generally Alces, supra note 24; Cohen, supra note 23.
a breach of the creditor’s duty to the guarantor;33 and so are the rules for contracting around the
default rules that these various legal regimes provide.34
As a result, the content of guaranty obligations depend primarily on the parties’ choice of
transactional form. A person guaranteeing a promise embodied in a promissory note can opt into the
rules of Article 3 by signing the note as accommodation party35, or into the rules of the common law
of suretyship by signing a separate written agreement. She can opt into Article 5 and the traditional
law of letters of credit by casting her guaranty in the form of a letter of credit, or into Article 9 by
offering specific property as collateral. Even within these broader transactional forms, furthermore,
she can vary her rights and duties by using particular terms of art, as in the Article 3 distinction
between guaranty of collection and guaranty of payment,36 or by providing specific conditions for
transfer and presentment in a letter of credit.37 She can, by issuing a limited rather than an unlimited
guaranty, mimic various risk-sharing devices common in insurance markets: including liability caps,
deductibles, co-payments.38 And, apart from a few exceptions, it is possible to contract around the
33. Compare, e.g., Restatement (Third) of Suretyship and Guaranty, §§ 37 and 42 (1996), with UCC §§3-605(e)
and (f).
34. Compare Restatement (Third) of Suretyship and Guaranty, §6 (1996) and UCC §3-605(I) (allowing guarantor
to waive defenses based on suretyship and impairment of collateral) with UCC §9-501(3) (providing that
certain rights of guarantor and duties of creditor with respect to disposition of collateral following default may
not be varied or waived in advance of default, and others cannot be varied or waived at all).
35. Of course, the note must also be negotiable for Article 3 to apply. Ronald Mann, Searching for Negotiability
in Payment and Credit Systems, 44 UCLA L. REV. 951 (1997), presents empirical evidence that such
negotiable promissory notes are relatively rare in modern business transactions.
36. See UCC §3-419(d).
37. See UCC §5-111 (providing that beneficiary, by transferring a letter of credit or demanding payment
thereunder warrants that all necessary conditions of the credit have been complied with.)
38. See Alces, supra note 24, at 457. The form such limits can take depends on the type of guaranty used. For
instance, if the guaranty takes the form of a put option on the underlying debt, the exercise price of the option
implicitly determines the deductible. A face-value exercise price is equivalent to no deductible; lower exercise
prices imply higher deductibles.
default rules of suretyship whatever the applicable legal regime.39 Indeed, such waivers of default are
the norm in many lending markets.40
To decide whether to cast one’s financial transactions in the form of a guaranty, which of the
various legal regimes of guaranty law to opt for if one does, and to decide whether to accept or to
disclaim the various default rules of each regime, therefore, it is necessary to understand the economic
purposes that guaranty contracts serve. In the next section, accordingly, I present an account of why
guaranties exist — why and when it is desirable to divide the functions of providing funds and of
bearing the residual risks of debtor default between two different parties, and why and when it is
desirable to use a three-corner conditional arrangement when doing so.
III. Guaranties and their alternatives: the choice of transactional form
In this section, I discuss why parties would or should want to use guaranties as part of their credit
transactions, as opposed to making alternative arrangements. The discussion focuses on the criterion
of economic efficiency — that is, of maximizing the net surplus, monetary or otherwise, that results
from the transaction. I do not attempt here to offer a rigorous defense of the efficiency criterion;
rather, I will merely observe that it is a primary and widely shared goal in the commercial context.41
In most business dealings the parties are motivated in substantial part by the prospect of economic
gain and have the opportunity to plan their arrangements deliberately and with the assistance of
professional advisers. Because the law also allows them substantial leeway to shape their contractual
39. See e.g., Restatement (Third) of Suretyship and Guaranty, §6 (1996); UCC §1–102. But see UCC §9-501(3)
for some possible exceptions. Whether the Article 9 limits on freedom of contract apply to guarantors as well
as primary obligors is an question on which courts have differed. [Citations to be provided.]
40. See Cohen, supra note 2, at 1042–43; RESTATEMENT (THIRD) OF SURETYSHIP AND GUARANTY, §48 (1996),
comment a; Lynn LoPucki et al., COMMERCIAL TRANSACTIONS: A SYSTEMS AND PROBLEMS APPROACH
(forthcoming 1998), at __.
41. For more general discussions of the efficiency criterion, see Jules Coleman, MARKETS, MORALS, AND THE
LAW (1988), ch.4 ; Richard Posner, THE ECONOMICS OF JUSTICE (1983), chs. 3 and 4; and the articles
appearing in Symposium on Efficiency as a Legal Concern, 8 HOFSTRA L. REV. 485 (1980), More specifically,
Louis Kaplow and Steven Shavell, Why the Legal System is Less Efficient than the Income Tax in
Redistributing Income, 23 J. LEGAL STUD. 667, argue that even if distributional equity is an important social
objective, it is more effectively promoted by using direct public instruments such as tax and transfer payments,
rather than through the rules of private law.
relationship, they accordingly have both incentive and ability to conserve on transaction costs. Thus,
an efficiency analysis operates in this setting as both an explanatory theory of commercial actors’
behavior, and as a normative benchmark by which to judge their efforts.42 The usual caveats and
admonitions to such an approach, of course, will apply here just as they do anywhere else.43
Before we begin the main discussion, however, two prefatory remarks are in order. First, though
I will often talk of the guarantor’s receiving an economic benefit from the guaranty transaction, the
analysis to follow applies equally well to guaranties that are uncompensated. In particular, the fact
that many guaranties are uncompensated and some are gratuitous does not diminish the importance
of the efficiency criterion, for persons with purely donative intent will ordinarily wish to make their
gifts in an efficient manner. In this regard, efficient giftgiving minimizes unnecessary transaction costs
while maximizing the value the beneficiary actually receives from a gift of given cost.44
In the case of a donative guaranty, more specifically, there are numerous ways to assist a person
in need of credit. One can lend the intended beneficiary funds at a below–market interest rate,
guarantee the beneficiary’s loan with another creditor, or subsidize the beneficiary’s interest
expenditures directly in an amount equal to the differential charged between high–risk and low–risk
loans. Depending on the particular circumstances of the beneficiary and the donor, these various
arrangements will have different costs and benefits, so that a guaranty may or may not be the most
42. Furthermore, so long as the transaction in question is an arms–length one, the parties have the option not to
enter into it, and they know the relevant economic risks and legal consequences, there are no obvious
distributional consequences from any change in the legal rules. As a general matter, the surplus from
exchange tends to be divided among contracting parties in proportion to their relative eagerness to enter into
the bargain. Any efficiency gains or losses resulting from a change in regime, accordingly, will be shared by
all. See generally Richard Craswell, Passing on the Costs of Legal Rules: Efficiency and Distribution in
Buyer-Seller Relationships, 43 STAN. L. REV. 361 (1991).
43. In addition to the sources cited supra in notes 41 and 42, see, e.g., A. Mitchell Polinsky, Economic Analysis as
a Potentially Defective Product: A Buyer's Guide to Posner's "Economic Analysis of Law", 87 HARV. L. REV.
1655 (1974) (discussing underlying assumptions of efficiency analysis); Arthur Okun, EQUALITY AND
EFFICIENCY: THE BIG TRADEOFF (1975) (discussing tension between efficiency and distributional equity);
Coleman, supra, note 41, (critiquing efficiency criterion from perspective of classical utilitarianism); Ronald
Dworkin, Is Wealth A Value?, 9 J. LEGAL STUD. 191 (1980) (critiquing efficiency from perspective of liberal
individualism); and the various critiques surveyed in Avery Wiener Katz, FOUNDATIONS OF THE ECONOMIC
APPROACH TO LAW (forthcoming 1998), ch. 8.
44. See generally Steven Shavell, An Economic Analysis of Altruism and Deferred Gifts, 20 J. LEGAL STUD. 401
(1991). Moreover, it is worth remembering that many formally uncompensated guaranties are motivated in
substance by ordinary considerations of self–interest. Cf. Richard Posner, Gratuitous Promises in Economics
and Law, 6 J. LEGAL STUD. 411 (1977) [explaining other apparently donative promises in such terms].
cost–effective form of assistance. As I will show below, the factors that determine whether a
guaranty is efficient in the context of a gift are analogous to those that determine its efficiency in the
context of a bargain.45
Second and more importantly, my focus on efficiency is not intended to suggest that individual
profit maximization always leads to an efficient outcome, or that actual guaranties are always
motivated by efficiency considerations. On the contrary, the standard litany of market failures —
scale economies, monopoly, externalities, and imperfect information — are just as relevant here as
in any other exchange setting. For instance, if guarantors underestimate the costs of making a
guaranty, as some gratuitous guarantors surely do, private contracting will lead to too many
guaranties and, conversely, too few interest subsidies or direct low–interest loans.46 Similarly, the
parties to a credit transaction may in many instances find otherwise inefficient arrangements to be
45. Admittedly, some of the social functions of giftgiving — for instance, the display of spontaneity, selflessness,
extravagant generosity, or affluence — are in apparent tension with narrow stereotypes of efficiency and
rational maximization. Compare, e.g., the accounts of giftgiving in Joel Waldfogel, The Deadweight Loss of
Christmas, 83 AM. ECON. REV. 1328 (1993) (arguing that cash gifts are more efficient than noncash because
they can be spent on whatever the recipient wishes, and calculating four billion dollars in annual deadweight
losses from noncash holiday gifts), with those of Richard Titmuss, THE GIFT RELATIONSHIP: FROM HUMAN
BLOOD TO SOCIAL POLICY (1970); and William I. Miller, HUMILIATION AND OTHER ESSAYS ON HONOR,
SOCIAL DISCOMFORT, AND VIOLENCE (1993), ch. 1. In many instances, however, and especially in those
situations mediated by the presence of a third party, the symbolic and expressive aspects of the gift relationship
can be accommodated without sacrificing the efficiency of the transaction — as the existence of entire legal
fields such as estate planning amply illustrates. Furthermore, as I suggest below (see text accompanying note
78, infra) some of the sociological and symbolic considerations that might favor the use of guaranties in the
gratuitous context — for example, not wishing to establish an arms–length business relationship between the
guarantor and the beneficiary — can be analogized to the more purely pecuniary transaction costs that animate
guaranties in a commercial setting.
46. Indeed, several important doctrines of suretyship law are motivated primarily by concern for the incautious or
noncalculating guarantor. These include the formal requirement of consideration, which has been interpreted
to have rather less bite in the commercial as opposed to the donative context [see Restatement (Third) of
Suretyship and Guaranty, §9 (1996)], the special Statute of Frauds for suretyships [see Restatement §11]; and
the greater procedural and substantive scope given to suretyship defenses in the donative context [see, e.g.,
Restatement §49, providing that compensated but not gratuitous guarantors bear the burden of proof with
regard to the extent of economic loss resulting from the creditor’s improper actions]. Modern consumer
protection law offers additional safeguards for casual guarantors [see, e.g., the FTC’s Credit Practices Rules,
especially those codified in 16 C.F.R. Part 444.3 (“Unfair or Deceptive Cosigner Practices”)]. And in
addition, many courts have shown special solicitude to the special form–contract and boilerplate problems
arising from guaranty contracts, leading those courts to interpret claims against guarantors strictly and
defenses broadly. See generally Peter A. Alces, The Efficacy of Guaranty Contracts in Sophisticated
Commercial Transactions, 61 N. CAROLINA L. REV. 655 (1983).
privately profitable because those arrangements allow the principals to extract rents from or to impose
externalities on third parties.
In the context of commercial credit, the other creditors of D and G, both present and future, are
obvious targets for such rent extraction. If such creditors are fully aware of the relationship among
C, D, and G, of course, and if the market for credit is competitive, they can fully protect themselves
by adjusting the terms of their contracts with D and G to compensate for any resultant changes in
risk. Many outside creditors, however, will be unaware of the details of the participants’
arrangements (as in the case of incidental trade creditors whose loans are too small to make it
worthwhile keeping track of D and G’s doings), unable to alter the terms of their loans in response
(as is the case with tort claimants and other involuntary creditors), or both (as with employees not
represented by collective bargaining who are creditors to the extent of unpaid wages and benefits).
Accordingly, these third parties may suffer negative externalities if the parties to the principal credit
transaction are allowed full freedom of contract.47
It is difficult, however, to know in which direction this factor cuts in the guaranty context.
Because (in contrast with secured credit and mortgages) there is no centralized system of notice or
recording for guaranties, and because it is difficult for parties not in contact with D to gauge his
potential for default, G’s other creditors may inadequately adjust for the increased risk the guaranty
poses for their claims. This would suggest an incentive for the issuance of too many guaranties. On
the other hand, because parties not in contact with G may be unaware of or unable to appreciate the
value of her guaranty, D’s and C’s other creditors are likely to adjust inadequately for the decrease
in risk the guaranty provides them. This countervailing factor would imply that too few guaranties
are being written.
Conversely, guaranties can also impose externalities on debtors with whom D is in competition
for credit. To the extent that a guaranty from G allows D to improve his credit prospects by signaling
his reliability, part of this gain comes at the expense of the general pool of credit applicants who share
his observable characteristics. D will now be perceived as more reliable than average, but the others
47. This argument should be familiar to followers of the policy literature on mortgages and personal property
security interests, where it has commonly been offered as a reason to limit the availability of and the priorities
arising from secured credit. See Lynn M. LoPucki, The Unsecured Creditor’s Bargain, 80 VA. L. REV. 1887
(1994), Lucien A. Bebchuk and Jesse M. Fried, The Uneasy Case for the Priority of Secured Claims in
Bankruptcy, 105 YALE L. J. 857 (1996), and Lynn M. LoPucki, The Death of Liability, 106 YALE L. J. 1
(1996).
in his risk pool will, as a direct consequence, be perceived as less reliable than average. Unless those
other applicants are able to obtain their own guaranties, accordingly, they will receive less credit on
less favorable terms than before. Additionally, those applicants who do go to the expense and trouble
of providing a guaranty will find that the more guaranties are provided, the more their signaling value
is diluted. The result, as in any market with costly signaling, may be a rat race in which all parties are
worse off than if none had tried to signal in the first place.48 The proper policy response to such
externalities if they exist, however, is not obvious; depending on the particular circumstances it may
be possible to improve social welfare either by subsidizing information acquisition or by taxing it.49
Various other commonly articulated rationales for guaranties can be understood in similar terms.
For instance, it is often claimed that guaranties are preferred for accounting reasons — that they look
better on one or more of the parties’ balance sheets than would an outright unguaranteed loan.50 This
claim, if true, implies a set of externalities precisely analogous to those we have just discussed, for
what it comes down to is that the accounting and auditing specialists who supervise the principal
parties for the benefit of third party creditors and investors use imperfect categories to assess and
communicate the parties’ financial health. If the combined financial statements of the principals
appear stronger when they use guaranties instead than loans, then third party creditors and investors
will be induced to deal with them on more favorable terms than they otherwise would. At the same
time, the financial statements of competing actors who do not use guaranties will look relatively weak
in comparison, reducing their credit prospects and increasing their borrowing costs. To the extent
48. The classic formulations of this argument can be found in A. Michael Spence, Job Market Signaling, 87 Q.J.
ECON. 355–74 (1973); see also George Akerlof, The Economics of Caste and of the Rat Race and Other
Woeful Tales, 90 Q.J. ECON. 599-617 (1976).
49. See, e.g., Severin Borenstein, The Economics of Costly Risk Sorting in Competitive Insurance Markets, 9
INT’L. REV. L. ECON. 25 (1989) [arguing that it may improve efficiency to prohibit insurers from spending
resources to distinguish high–risk from low–risk insureds.]
50. For instance, under current rules of the Financial Accounting Standards Board (FASB), standby letters of
credit need be disclosed only in the form of a footnote to the balance sheet, rather than being listed explicitly
as a contingent liability. See, e.g., Medhat Helmi and Nitham Hindi, Raining on Banking’s Parade: New
Rules Promote Fuller Accounting of Liabilities, 40 NATL. PUB. ACCOUNTANT 17 (1995) [advocating that
FASB rules be amended to treat standbys as contingent liabilities, and offering empirical evidence to
demonstrate that sufficient data exists to estimate the value of such contingencies].Attempts to correct for the
imperfections of such accounting practices are ongoing. Under the risk–based capital standards established in
1988 by the Basle Committee on Banking Regulation and Supervisory Practice and still in the process of
implementation through various federal regulations, for example, most domestic banks will be required to
include letters of credit when calculating their risk–based capital ratios. See Pawlowic, supra, note 13, at
416–417.
the apparent improvement on paper exceeds the real economic gains of the transaction, however, any
benefit to the principals comes partly at these third parties’ expense.51
Guaranties can also be driven by tax considerations, since under the Internal Revenue Code the
same transaction is taxed differently depending on whether it is structured as a guaranty, a loan, or
the sale of a put option.52 For instance, the interest payable on a loan that extends over multiple years
is taxed under the original issue discount rules, resulting in a relatively even pattern of deductions for
D and income receipts for C over the life of the loan even where the actual interest payments are
front–loaded. If the loan is guaranteed, in contrast, a portion of the interest payments are converted
into suretyship fees, which can be fully expensed by D (and included as income for G) at the time they
are paid.53 If D is in a higher marginal tax bracket than C, or if G is in a lower bracket than C,
accelerating the realization of these matched payments has the effect of lowering the total present
value of taxes paid. Various other tax advantages, and disadvantages, of guaranties leave ample room
for tax arbitrage in the design of credit transactions. The reduction in the parties’ combined tax
liability from such arbitrage, however, is effectively an externality imposed on the Treasury, and
ultimately on the taxpaying public.
Finally, the parties may cast their relationship in the form of a guaranty in order to evade the
substantive constraints of other formal or informal regulations. One straightforward example of such
regulation would be the practical constraints imposed by public opinion regarding U.S. direct
government subsidies to Israel or Mexico, which under current budgetary rules would require
accompanying tax increases or cuts in other more popular programs. Federal loan guaranties to these
countries, in contrast, have until been recently treated as entirely off–budget, and thus better able to
escape both public and Congressional scrutiny.54 A second example, less obvious but commercially
quite significant, is provided by the reserve requirements imposed by the Federal Reserve System and
51. See, e.g., Elijah Brewer and G. D. Koppenhaver, The Impact of Standby Letters of Credit on Bank Risk: A
Note, 16 J. BANKING & FIN. 1037 (1992) [presenting empirical evidence that equity markets accurately
perceive and price the risk of standby letter of credit issuance].
52. See generally David S. Miller, Federal Income Tax Consequences of Guaranties: A Comprehensive
Framework for Analysis, 48 TAX LAWYER 103 (1994).
53. Id., at __.
54. See, e.g., Kate Stith, Congress' Power of the Purse, 97 YALE L.J. 1343, 1379 (1988) [discussing loan
guaranties as one of several “backdoor” appropriations mechanisms]; Sanford, supra, note 22, at 377– 380
[discussing effect of recent budgetary reforms of as they affect such guaranties].
the lending limits imposed by the Comptroller of the Currency on federally chartered banks, both of
which are intended to contain the risk of bank failures, to encourage depositor confidence, and to
control the money supply as a whole. Under the Fed’s requirements, for instance, banks must hold
a certain percentage of their liabilities on reserve in the form of cash or deposits with one of the
Federal Reserve Banks; the effective result is that the amount of credit extended by any bank is
limited to a given multiple of the reserves it keeps on hand.55 Because Fed deposits earn a lower
interest rate than commercial loans, however, and because cash earns no interest at all, banks
accordingly have a strong incentive to lend their reserves out to the limit. Until the late 1980's,
however, the Fed and the Controller failed to count guaranty obligations incurred under standby
letters of credit against banks’ lending limits, on the theory that these obligations were merely
contingent.56 These loopholes have been credited with encouraging the explosion of standby letters
of credit in the 1980's, as banks constrained not to make additional direct loans found that they could
increase their effective lending by casting a portion of such loans in the form of guaranties.57
Notwithstanding these important non–efficiency explanations for guaranties, I abstract from them
in the following discussion. I do this for three related reasons. First, an efficiency analysis is
interesting in itself, and useful to both private and public actors planning transactions. Even if we
grant that in the current political environment the way foreign aid is packaged for public consumption
will always outweigh efficiency considerations, for instance, it is still important to know whether it
is cheaper to provide such aid in one form rather than another. Second, a clear understanding of
guaranties in efficiency terms is analytically prior to addressing any externalities and regulatory
inefficiencies they may produce. In order to decide how guaranties should be taxed, for instance, or
whether and to what extent they should be subject to bank reserve requirements, we need to know
what they are and how they differ functionally from the other transactions for which they are a
55. See generally Lawrence S Ritter and William Silber, PRINCIPLES OF MONEY, BANKING, AND FINANCIAL
MARKETS (1991), at __.
56. The loopholes were largely though not entirely closed by 12 C.F.R. §§32 and 208.8d (1990), which subjected
standby letters, as appropriately discounted for risk, to the same lending regulations as ordinary loans.
Because the categories for discounting such risk are imperfect, of course, some minor regulatory incentive to
substitute standbys for direct loans still remains.
57. See, e.g., Barbara Bennett, Off-Balance Sheet Risk in Banking: The Case of Standby Letters of Credit,
FEDERAL RESERVE BANK OF SAN FRANCISCO ECONOMIC REVIEW 19 (Winter 1986), at 22–23 [but concluding
that the primary reason for growth in the use of standbys stemmed from real economic demand for guaranties
rather than capital regulations ].
Figure 2. Alternative configurations of credit
substitute. Third, the non–efficiency reasons for guaranties are local and contextual; they vary in
detail from industry to industry and regulatory regime to regulatory regime. The basic economic
efficiencies underlying guaranties, however, are general and universal. A relatively short and simple
set of principles suffice to explain analogous relationships across a wide and disparate number of legal
and policy areas, allowing insights gleaned from one field to inform our understanding of all the
others.
I begin, therefore, by considering the main alternatives to guaranty transactions, and the costs and
benefits these alternatives produce. I then turn to a comparison of the alternatives.
A. Guaranties, unguaranteed lending, and intermediation
Consider the situation in which a debtor D seeks credit from two potential creditors, C and G.
There are several ways to provide funds to D, the three main types of which are depicted in Figure
2, where arrows indicate the flow of credit. First, D could borrow from C alone, leaving G out of
the transaction entirely; this possibility is depicted in part (a) of the figure. Second, D could borrow
from C and supplement it with a guaranty from G. This possibility is depicted in part (b) of the figure;
the dotted arrows indicate that the extension of credit is contingent rather than unconditional.
Specifically, if D does not repay the primary loan, then G will become indebted to C and D will
become indebted to G. Third, G could borrow from C and independently lend to D, a possibility
depicted in part (c) of the figure. In this case both obligations are unconditional rather than
contingent. C would then have no direct claim against D, though G’s claim against D would be one
of the assets to which C could ultimately look for satisfaction if G falls into default with C.58
For each of these three possibilities there exists an alternative and symmetric arrangement with
G substituting for C and vice versa. For instance, D could borrow the entire amount from G rather
than from C, and so on. Additionally, the parties could provide part of the desired funds through one
form and the rest through another, as when C and G separately and independently lend D half of what
he requests. These further possibilities are ignored in Figure 2 for the sake of simplicity, but they are
discussed below where relevant.
The precise costs of credit will vary depending on the form of the transaction. In the simple case
where C lends alone, there are three main components of credit cost. The first is the marginal cost
that C bears in obtaining liquid funds to lend. This cost reflects the explicit or implicit interest rate,
iC , that C must pay for such funds when borrowing, whether by drawing down its own savings, by
reducing loans to other borrowers, or by engaging in financial intermediation with other potential
creditors. It also reflects C’s individual cost of liquidity, lC , which should be understood here in a
Coasian sense; as I use the term it denotes the transaction costs that C must incur to convert its assets
into a readily acceptable medium of exchange.59 If, for example, all of C’s free assets happened to
be tied up in a parcel of real estate, its liquidity costs would include all the expenditures and
opportunity costs associated with mortgaging or selling the parcel, including commissions, appraisals,
surveys, title insurance, transfer taxes, attorneys’ fees, and the like. Such liquidity costs will of course
depend on the particular form in which C’s assets are held. If C’s assets all take the form of
certificates of deposit, in contrast, its liquidity costs may be limited to the interest penalties it incurs
for early withdrawal; and if C’s entire wealth is held in the form of cash in a mattress, its liquidity cost
will be close to zero (though in both cases it is still necessary to consider, in an amount discounted
58. Under §541 of the U.S. federal bankruptcy code, 11 U.S.C. §§101 et seq., all of a debtor’s legal or equitable
property interests as of the commencement of bankruptcy, including unpaid accounts receivable, become
property of the bankruptcy estate, to be collected and managed by the trustee for the benefit of creditors. In
addition, the trustee has the discretion under §365 to assume any executory contract rights of the debtor.
59. See generally James Tobin, Liquidity Preference as Behavior Towards Risk, 25 REV. ECON. STUD. 65 (1958).
for the probability of its occurrence, the additional transaction costs that C will have to incur if it
loans out all its cash and has to borrow more funds in an emergency). Both of these determinants of
C’s cost of funds, interest and liquidity, will in general depend on the amount of credit being
extended, but to simplify the discussion I will speak as if total credit costs are strictly proportional
to the amount being lent.
The second main component of C’s credit cost is its transaction costs of dealing with D. These
include the ordinary incidents of processing a loan: underwriting, time spent by loan officers, clerks,
and other administrators, the costs of drafting and executing a loan contract, and servicing the loan
after it is made. They also include the costs of investigating D’s creditworthiness and auditing his
behavior over the life of the loan Such monitoring costs will depend on the amount being lent, but
more importantly, they will also turn on the precise nature of C’s relationship with D. If C has never
dealt with D before, it will need to spend time acquainting itself with D’s business affairs and credit
history, with the merits of any projects D plans to pursue the funds being borrowed, with other
competing demands on D’s resources, and with similar matters, before making any loan. These costs
will be lower if C has dealt with other persons in D’s position in the past, and lower still if C has dealt
with D personally. Conversely, if C plans to deal with D again in the future these costs can be
amortized over multiple occasions, whereas if this is a one–shot transaction they will have to be
recovered all at once.60 Similar considerations apply to the costs of supervising the loan after it has
been made (since a rational creditor will want to keep tabs on its debtor to make sure he is using the
funds appropriately), taking adequate precautions to ensure D’s ability to repay it, and maintaining
the value of any necessary collateral. To emphasize the relational aspects of such monitoring costs,
I denote them here with the symbol mCD.
Third, even with optimal monitoring there still will remain some risk of debtor insolvency and
default. The resultant loss to be expected depends both on the probability of such a default, and on
the amount of funds that can be recovered afterwards. Both these quantities will depend on the
amount of investigation and auditing that C has done, on C and D’s individual characteristics, and on
external factors out of both parties’ control. Furthermore, because C must agree to accept this risk
if it is to lend at all, C’s subjective assessments of risk, based on past experience with D and other
60. See generally Robert E. Scott, A Relational Theory of Secured Financing, 86 COLUM. L. REV. 901 (1986)
[emphasizing the importance of fixed investigative and monitoring costs in the credit context, and arguing that
such costs explain why many individual debtor–creditor relationships tend to be long–term.]
debtors who appear in C’s eyes to be similar, are also relevant. Again, to highlight the relational
aspect of such considerations, I denote the monetary value of this risk, as perceived by C, as rCD.
The total cost of credit under an unconditional loan from C to D is thus iC + lC + mCD + rCD. In
a competitive credit market — that is, one in which C contends with many other potential creditors
for D’s business — competition among creditors will bid price down to marginal cost, so that this
also is the amount that D will have to pay for credit. A competitive market, of course, does not
ensure that the cost of credit is low. If D has little experience with borrowing, or if he resembles
other debtors whose risk of default have been high in the past, or if the market’s subjective opinion
of him are simply unfavorable, then the competitive price of credit he faces will be high, for the
expected monitoring and default risk costs of creditors who deal with him will also be high.
Furthermore, if he faces a non–competitive credit market, he will probably have to pay an additional
mark–up above and beyond marginal cost. Still, there is no particular reason to think that the amount
of any non–competitive mark–up will depend on the nonprice terms of the loan, or that lender market
power will result in an inefficient mix of auditing, investigation, or other terms.61
Conversely, if D borrows directly and unconditionally from G, his expected price of credit will
equal the quantity iG + lG + mGD + rGD — the sum of G’s costs of interest, liquidity, investigation and
monitoring, and subjectively anticipated default risk when dealing with D on such a basis. In either
case, D must also add to the market price of credit his own personal transaction costs from
borrowing, which include not just time, trouble, and the expense of dealing with legal papers, but the
costs of investigating and monitoring C and G in turn in order to ensure that they do not themselves
act in an opportunistic or negligent fashion. The size of these transaction costs will also depend on
the nature of D’s relationships with the other parties, and can be respectively denoted as tDC and tDG.
When choosing between G and C as potential direct creditors, accordingly, D will prefer to borrow
where these total costs are least, other things being equal.
B. The cost of credit in a guaranty transaction
61. In general, while noncompetitive markets tend to produce inefficiently high prices (that is, prices in excess of
marginal costs), they have no tendency toward inefficient nonprice terms. On the contrary, a profit–seeking
monopolist or oligopolist will ordinarily want to choose the nonprice terms that best suit the preferences of the
marginal consumer, for this maximizes the markup he or she can charge. For fuller explanations of this point,
see A. Michael Spence, Monopoly, Quality, and Regulation, 6 BELL J. ECON. 407 (1975),William Comanor,
Vertical Price–fixing, Vertical Market Restrictions, and the New Antitrust Policy, 98 HARV. L. REV. 983
(1985), and Richard Craswell, Passing on the Costs of Legal Rules: Efficiency and Distribution in
Buyer–seller Relationships, STANFORD L. REV. 361 (1991).
The addition of a guarantor to the transaction unambiguously reduces C’s costs. This is so for
several reasons. First, because C can anticipate that G will rationally wish to engage in some
monitoring of D, C can reduce its own monitoring efforts to a corresponding extent. Second, even
if G engages in no monitoring, her willingness to underwrite D’s debt credibly signals her subjective
belief that D is relatively likely to perform. To the extent that C has reason to think G’s risk
assessments accurate, this signal lowers its own expected costs of monitoring and default. As an
illustration, consider a mother with two sons, one industrious and one irresponsible, both of whom
are seeking credit. Because the mother knows that the former son is less likely to default than the
latter, the expected cost to her of giving him a guaranty is lower. Other things being equal,
accordingly, she is likelier to guarantee the former’s debt than the latter’s.62
Third, to the extent that D values his relationship with G, cares about G’s welfare, or anticipates
that G will be better able to collect than will C alone, D has incentives to take greater precautions
against default. And fourth and most obviously, should D become insolvent, C will have a backup
source of funds to pursue. Except in the extreme case where D’s and G’s assets are essentially the
same (as with a parent and subsidiary corporation that operate de facto as a single entity), the chance
that both D and G will become insolvent simultaneously is an order of magnitude less than the chance
that either one will become insolvent alone.63
As a result, C would plainly prefer adding G to the transaction if there were no costs associated
with doing so. But of course there is a cost — the monitoring and risk costs that will be imposed on
G in her role as guarantor. To the extent that C is relieved from the responsibility for monitoring, G
is burdened with it. To the extent that C can recover from G following D’s default, G (and G’s other
creditors) are at risk. Thus, many, though not all, of the costs saved by C on a guaranteed loan are
simply shifted to G.
The total costs of a guaranteed loan can thus be calculated as follows. First, we start with C’s
costs of interest, liquidity, monitoring, and default risk: iC + lC + mCD|G + rCD|G. (The notation CD|G
62. Unless she feels that the industrious son, being better able to take care of himself, is in less need of her
largesse. But even so, the spendthrift’s default risk to creditors makes a guaranty a relatively costly way of
lending him assistance.
63. This last factor explains why the wealth, financial strength, and credit reputation of the guarantor matters to
lenders in actual guaranty transactions. Additionally, to the extent that G turns out to be de facto uncollectible
after the fact, her incentives to monitor the debtor or accurately to signal his default risk are attenuated. On
the issue of guarantor financial strength, see also subsection D, infra.
denotes C’s cost of dealing with D given that G is also in the transaction; it also reflects any
monitoring that C wishes to do of G.) C’s interest and liquidity costs are unchanged by G’s presence
in the transaction, but its combined monitoring and default risk costs are lower for reasons explained
above: mCD|G + rCD|G < mCD + rCD .
Next, we must add in the expected costs that G incurs in her role as guarantor; these costs are
of three types and differ in amount from the costs that G expends when making a direct loan. First,
while G need not pay interest costs in order to provide a guaranty, she does incur some monitoring
costs. Second, she also will perceive some risk of default, the cost of which will depend on her own
assessment of D’s situation, the investigative efforts she undertakes, and her ability to collect from
D. Both of these costs, furthermore, are influenced by C’s involvement in the transaction. The
influence can cut one of two ways. In the event that C’s monitoring and investigation decreases D’s
chances of default, this will lower G’s monitoring and default costs; on the other hand, if C behaves
opportunistically toward D or toward the collateral, this may increase G’s costs. (Indeed, for this
reason G may wish to engage in some investigation and monitoring of C as well as of D.)
Accordingly, G’s monitoring and default costs under a guaranty will depend on her relationship with
C, and are denoted here as mGD|C and rGD|C.
Third and finally, in the event of a default G not only must pay out on her guaranty, she must also
raise the necessary funds to do so; this will force her to incur personal liquidity costs at that time.
Such liquidity costs need not be paid up front, however, and can be discounted by the probability of
a default, rGD|C. Thus, G’s total expected costs from a guaranty will equal mGD|C + rGD|C (1 + lG).
In a competitive credit market, D will have to pay the sum total of C’s costs and G’s costs in
order to obtain a guaranteed loan. 64 On top of that C must consider his own transaction costs of
dealing with the creditor and guarantor, both of which will depend on his prior relationship with those
parties. The total costs of dealing with both C and G together, moreover, are likely to be higher than
the individual costs of dealing with either one alone under a direct unconditional loan, since there are
now two persons to investigate and monitor and two persons to whom C owes a duty of care.
Accordingly, we denote these transaction costs as tDC|G + tDG|C.
64. This is so even if the guaranty is a gratuitous one, for G’s ability and inclination to make gifts to D is
presumably finite; any costs she incurs in making the guaranty will reduce the funds she has available for
alternative and future gifts. Thus, D must count the direct costs to G as an approximation of the opportunity
cost to him resulting from such a reduction.
Without resorting to a great deal of tedious algebra, it should be apparent that a loan made by C
and guaranteed by G is cheaper than a simple unguaranteed loan from C if, and only if, the expected
savings achieved by shifting monitoring, investigation, and risk default costs from C to G outweigh
the increased transaction costs — in terms of monitoring, investigation, and liquidity — that arise
from having to maintain three relationships among the parties instead of just one. In short, guaranties
are profitable because the guarantor, due to a sufficiently great comparative advantage in
investigating, supervising, and collecting from the debtor in the event of default, is the least–cost
monitor. The creditor pays the guarantor to do this monitoring, either explicitly or implicitly, and the
guarantor either keeps the proceeds as her compensation or forwards them to the debtor as a gift.
This least–cost monitoring explanation of guaranties fits well with the stereotypical examples
discussed above. Consider first the case of interfamilial guaranties. Parents generally know their
children better than third party lenders do. They are ordinarily familiar with the children’s past
financial behavior through their previous financial and emotional relationship; in most cases, they have
relatively more accurate information about the children’s day–to–day doings. Because of the
awkwardness and guilt most adult feel when being in debt to one’s parents, furthermore, and because
of the utility they attach to their parents’ welfare, the children will work harder to avoid defaulting
when the parents are potentially liable for filial debts. The parents are also better able to collect from
their children in the event of default, not only because they have better information about hidden
assets, but because the perceived moral obligation to repay a family member for a bad loan is likely
to survive bankruptcy where a commercial obligation does not, and because ultimately parents have
the option of collecting via a reduction in future gifts and bequests.65
Similarly, controlling shareholders hold a comparative advantage in investigating and supervising

the corporations they oversee. As residual claimants, they enjoy the marginal benefits and suffer the
marginal costs of the company’s actions, and hold the balance of power for certain critical decisions.
They may even be entitled to dissolve the corporation entirely. All these considerations make them
good candidates to guarantee corporate debts. Parent companies have similar comparative
advantages in monitoring their subsidiaries, and sibling companies their siblings. Subsidiary
companies may even have a comparative advantage in monitoring their corporate parents, especially
when there is common or overlapping management.
65. On all these considerations, and for a general discussion of implicit financial contracts between parents and
children, see Gary Becker and Kevin Murphy, The Family and the State, 31 J. LAW & ECON. 1 (1988).
In both the corporate and familial contexts, furthermore, such guaranties also serve as a
safeguard against opportunistic behavior by the guarantors. For instance, because of limited liability,
creditors of corporate debtors need to worry not only about whether the underlying project being
financed is sound. They also must worry about whether the proceeds of the loan will remain within
the corporation, as opposed to being funneled to a controlling owner, related entity, or other insider.
Minimum capital requirements, the corporate law duties of care and loyalty, fraudulent conveyance
law, and the doctrine of piercing the corporate veil all substantially help to police such misbehavior.66
In addition, however, many lenders insist on receiving guaranties from those actors deemed most
subject to temptation. Similarly, a lender contemplating a loan to an individual borrower will often
insist on receiving a guaranty from the borrower’s spouse, especially if the spouse holds a shared
interest in property pledged as collateral or otherwise holds the power to disrupt attempts at
collection. In these contexts, therefore, monitoring should be understood to include not just
affirmative efforts to reduce the risk of default, but refraining from moral hazard and opportunism.
The general practice of standby letters of credit can also be understood in such terms.
Commercial banks will have a comparative advantage in monitoring their existing credit customers,
as their prior relationships with such customers allows them to obtain and evaluate relevant
information at low cost. They also have many more ways to influence such debtors’ behavior after
a loan is made. For instance, they can call in, refuse to renew, or exercise covenants on separate
loans. In addition, they may have established general “floating liens” on the debtor’s property — that
is, mortgages or security interests that cover after–acquired property or that provide that existing
collateral will also serve to secure future advances.67 Such cross–collateralization clauses improve
the bank’s ability to collect on any advances it makes pursuant to a guaranty; they also give the bank
substantial leverage over the debtor by making it difficult for him to engage in subsequent borrowing
66. Guaranties by subsidiaries of their parent companies pose some special legal problems in this regard, however,
since the owners of the parent have a somewhat diluted incentive to ensure that the subsidiary receives fair
value for the parent’s guaranty. In the event that the subsidiary becomes insolvent, accordingly, its creditors
are likely to attack the guaranty as a fraudulent conveyance or an impairment of required capital. On the
problems raised by such “upstream” guaranties, see William H. Coquillette, Guaranty of and Security for the
Debt of a Parent Corporation by a Subsidiary Corporation, 30 CASE WESTERN L. REV.433 (1980).
67. Under U.C.C. §9–204, for instance, parties to a security agreement may provide that collateral acquired at any
time secures advances whenever made, including subsequent advances made when making good on a
guaranty. With limited exceptions [i.e., §§9–301(4), 9–307(3), 9–312(3), (4), and (7)], a security interest
based on such future advances will have the same priority over competing creditors as the original security
interest created by such an agreement.
without the bank’s permission.68 Perhaps most importantly, the prior relationship gives the bank the
power to ruin the customer’s reputation with other lenders. Since an unfavorable credit reference
from a longstanding creditor carries great weight in underwriting decisions, the bank can credibly
threaten to raise the customer’s cost of borrowing for the foreseeable future; such a threat serves as
a powerful incentive for D to repay.69 For all these reasons, then, bank guaranties can substantially
cut the expected costs of monitoring and default relative to unguaranteed loans.
Construction suretyships and payment bonds have a similar underlying motive.70 A general
contractor overseeing a construction project typically obtains significant advances from the project’s
overall sponsor and small–scale trade credit from a variety of project participants: subcontractors,
mechanics, materialmen. Rather than having each of these participants undertake the costs of
investigating the contractor’s credit and monitoring his use of funds, however, it is far cheaper to hire
a professional surety company to do the job and bear the associated risk of default. This arrangement
saves on duplicative monitoring effort and better allows the various trade creditors to focus on their
respective comparative advantages in plumbing, paving, and the like, enabling them to submit lower
bids initially. The surety, due to its experience in the industry, is also likely to be more effective than
the sponsor at picking up the pieces of the project and seeing it to conclusion should the contractor
withdraw. Using a surety, accordingly, lowers the expected cost of project completion by enough
to cover the required payment bond.
Finally, the example of joint and several liability among apartment co–tenants also fits our generic
explanation. Apartment roommates have an independent and ongoing incentive to investigate and
supervise each others’ reliability. They regularly extend each other petty credit in connection with
shared household expenses, and more importantly, share access to each others’ personal property.
They are also privy to each others’ daily comings and goings. With regard to claims such as damage
68. See Ronald Mann, The Role of Secured Credit in Small–Business Lending, 86 GEO. L.J. 1 (1997) [arguing
and presenting empirical evidence that limiting subsequent borrowing by the debtor is the most important
motivation for secured credit in actual practice.]
69. On the value of ruining a reputation as a contractual enforcement device, see Lisa Bernstein, Opting Out of the
Legal System: Extralegal Contractual Relations in the Diamond Industry, 21 J. LEGAL STUD. 115, __ (1992).
70. Thanks to Jim Bowers for suggesting this example and a good part of its analysis. Similar considerations
apply in other businesses besides construction. See, e.g., Mark C. Phillips, The Role of Completion Bonding
Companies in Independent Production, 12 LOY. L.A. ENT. L.J. 97 (1992) [describing role of guarantors in
motion picture industry ].
to common areas, furthermore, the landlord is typically unable to determine individual responsibility
at reasonable cost. It makes sense, therefore, for landlords to hire roommates to monitor each other
by providing for joint and several liability rather than contracting for several liability alone — at least
up to the liquidated amount of a security deposit.
C. Why does the guarantor guarantee rather than simply lend?
The foregoing account of guaranties is straightforward and intuitive, but it is not the whole story.
Why, if guarantors have a comparative advantage at investigation, supervision, and collection, don’t
they just make loans themselves, thus saving the extra transaction costs of a three–corner
relationship? Or to take a specific example, why do commercial banks issue standby letters of credit
to their customers, instead of directly lending to them? The answer cannot be that they wish to limit
their exposure, for under the terms of the standby letter they are at risk for the customer’s entire debt
in the event of default. The banks’ desire to exploit the previously mentioned loopholes in federal
lending regulations may partly explain their use of standby letters in the early 1980's, but today those
loopholes have largely (though not entirely) been closed.71 Is there an efficiency explanation?
The answer to this question lies in recognizing that while monitoring and investigation are part
of the cost of credit, they are not its entire cost. Recall that G’s credit costs as a direct lender also
include its implicit interest rate iG and its liquidity cost lG . As a guarantor, however, G avoids paying
for implicit interest, and it only need pay for liquidity in the event of a default. Interest and liquidity
costs do not disappear if G acts as guarantor, of course; they are simply transferred to the primary
creditor C. But if C’s cost of liquid funds is sufficiently lower than G’s, it can more than make up
for the extra information and transaction costs incurred by a three–corner arrangement.
The point is easily demonstrated with a numerical example. Consider the situation of a debtor
D who applies for a $10,000 loan from potential creditor C, to be repaid after one year. Suppose for
purposes of the example that the risk–free interest rate is 4%, reflecting the going rate of return on
one–year Treasury bonds, and that C has ready access to liquid assets at no additional cost beyond
this risk–free rate.72 Suppose also that if C lends to D, it will incur monitoring costs equal to 1% of
71. See notes 55–57, supra, and accompanying text.
72. For simplicity, we might imagine that C finances the loan to D by reducing its purchases of government bonds
by $10,000. More generally, as the preceding discussion makes plain, all market interest rates can be
decomposed into a component attributable to pure time value, a component attributable to liquidity, a
the face value of the loan, and that even after such monitoring, there will remain a 15% chance of
default. In this case, C will insist on charging a risk-adjusted interest rate of 20% [= 4% + 1% +
15%] if it is to make the loan. Depending on the project D is planning to finance, such a high cost
of funds may prevent the transaction from going forward; and even if D is willing and able to pay C’s
required rate of return, it may exceed the price ceiling established by the local usury laws.73
Alternatively, suppose that if D borrows from G instead of from C, G’s superior ability to monitor
and collect from D will cut the expected cost of default to 10%. Adding this default risk to the
risk–free interest rate of 4% and monitoring costs of 1%, this makes G’s cost of liquid credit 15%.74
But G holds her assets in relatively illiquid form, and thus bears a further 10% cost in making them
available for use by D. This raises G’s required interest rate to 25%, making her a less desirable
lender both from D’s viewpoint and from the social viewpoint.
If C makes the loan and G guarantees it, however, C’s access to liquid funds will keep the direct
cost of the loan down to 4%, and G’s superior monitoring ability will keep the cost of default down
to 10%. Of course, G will have to come up with liquid funds in the event of a default, but that event
occurs only 10% of the time. Thus while G’s liquidity costs under a direct loan are 10%, her
expected liquidity costs under a guaranty are equal to 10% of 10%, or 1%. Interest costs of 4% plus
monitoring costs of 1% plus default costs of 10% plus liquidity costs of 1% equal 16%. Of course,
there will be additional costs of structuring the transaction as a guaranty, including the costs of
duplicative monitoring by C and G. If we denote these additional transaction costs as t, then the total
cost of credit under a guaranty becomes 16% + t. If t is low enough, accordingly, a guaranty will be
cheaper and more efficient than a direct loan from either C or G.
A guaranty, in short, is a credit transaction in which one party — the guarantor — specializes in
information and monitoring, and another — the primary creditor — specializes in obtaining loanable
component attributable to risk, and a component attributable to transaction costs. The precise assumption
necessary here is that C has no additional cost of liquidity above and beyond whatever liquidity component is
implicit in the risk–free market rate.
73. See, e.g., Ark. Const. Art. 19, § 13 [maximum lawful interest rate equal to 5% above the Federal Reserve
discount rate]; Conn. Gen Stat. Ann. § 37-4 (West 1997) [maximum rate of 12%]; Fla. Stat. Ann. § 687.03
(West 1997) [maximum rate of 18% on debts less than $500,000].
74. To keep the calculations simple, the example assumes that G’s comparative advantage in monitoring is
reflected only in a reduced probability of default. The argument would be analogous, though more complex, in
the case where her out–of–pocket costs of monitoring is also lower than C’s.
funds. All of the guaranty arrangements described above accord with this basic principle. Let us
examine a few of them as illustrations.
Consider, for instance, our original example of the standby letter of credit. At any given moment,
the marginal cost of interest and liquidity varies among large commercial banks. Such banks have
different mixtures of long–term and short–term loans in their asset portfolios, and different risk
exposures based on the specific commercial activities of their customers and on the particular
investment products they have decided to back. It therefore makes sense for a bank with relatively
heavy exposure in the area of commercial real estate, for instance, to decide based on that position
that its marginal cost of liquidity is too high to justify a new loan to a longstanding client at market
rates. If the client can obtain liquid funds elsewhere, however, the bank can trade on its informational
investment in him by issuing a guaranty in his favor in the form of a standby letter of credit.
Conversely, another bank with temporarily superior access to a pool of liquid funds due, for instance,
to a special relationship with an institutional depositor, may find it worthwhile to make a loan to an
untried applicant if that applicant’s current lender is willing to provide a guaranty. The standby letter
of credit thus allows the borrower to benefit from the low monitoring cost that comes from
establishing a long–term financing relationship, while still retaining the option of shopping around for
the best pure price on liquid funds.75
Similarly, consider how our basic explanation operates in the context of a recourse sale of
accounts receivable. Under this arrangement, a retailer sells receivables to an account financer or
factor, promising to repay the factor for any accounts that turn out to be uncollectible.76 In the
framework of our model, the factor takes the role of the creditor, the retailer takes the role of
guarantor, and the consumers take the role of debtor. Asking why the guarantor doesn’t just lend
in this context is equivalent to asking why the retailer doesn’t just keep her accounts instead of
factoring them. The answer, apparent to anyone familiar with the business of factoring, is that the
75. In addition, the distinctive risk profile of a guaranteed loan may give it positive diversification value when
combined with the rest of the issuer’s lending portfolio. See, e.g., M. Kabir Hassan, An Empirical Analysis of
Bank Standby Letters of Credit Risk, 2 REV. FIN. ECON. 31 (1992) [presenting evidence of such value for
largest US banks and holding companies]; contra, Brewer and Koppenhaver, supra, note 51 [presenting
empirical evidence that the overall risk of issuing standby letters of credit is similar to the risk of balance sheet
lending.]
76. Or alternatively, the factor could make a secured loan to the retailer with the accounts serving as collateral.
With minor exceptions, these two arrangements are legally and functionally equivalent. See U.C.C.
§9–102(1)(b) and associated comment 2.
retailer wants the money now instead of later. She wants liquid funds with which to pay current
expenses and restock inventory, and this is her cheapest method of borrowing to obtain them. The
factor, who has superior access to loanable funds, provides her with those assets, in exchange for the
discount it gets on the accounts. But because the retailer is familiar with her own customer base, and
because the value of the accounts depends on the future services she provides, she is the least–cost
monitor, and thus appropriately bears the risk of consumer nonpayment.77
The typical circumstances surrounding the standard interfamilial guaranty also comport with our
theory. A parent may be highly confident of her child’s ability to repay a loan and have quite effective
means at her disposal to ensure that he actually does repay. As a result, she is willing to guarantee
that loan while receiving relatively little in exchange; indeed, the expected cost to her may be low
enough to justify making her guaranty a gift. But because the family assets are tied up in the house
and in the parents’ retirement savings accounts, making the same loan out of pocket would be quite
expensive. If the child should default, of course, the IRA’s will have to be cashed in and the house
mortgaged; but the chance of this happening is low — low enough to be outweighed by the benefits
conferred on the child.
Our model is also broad enough to encompass at least one plausible nonpecuniary explanation for
guaranties within the family. It has often been claimed that arms–length economic transactions
between friends and family members weaken the bonds of friendship and clan loyalty. For instance,
Steven Kelman has suggested that market interactions can lead to a “feeling–falloff” effect that lead
people to devalue their personal relationships with those with whom they deal; and some feminist
critics of law and economics, notably including Frances Olson and Margaret Radin, have argued that
economic analysis (and economic analysis of family law in particular) tends to commodify family
relations, undermining the love and altruism that are essential to the maintenance of household
cooperation and the healthy upbringing of children.78 If such claims are true, and if a direct
77. If she were not the cheaper monitor, conversely, the sale would be on a nonrecourse basis, not a recourse one.
The factor would then act as a direct lender, bearing the risk of default as well as providing liquid funds. See
generally Janet Kiholm Smith and Christjahn Schnucker, An Empirical Examination of Organizational
Structure: The Economics of the Factoring Decision, 1 J. CORP. FIN. 119 (1994) [presenting evidence that
factors are more likely to be used to finance trade credit, ceteris paribus, when the seller’s information and
monitoring costs are high].
78. See Steven Kelman, WHAT PRICE INCENTIVES: ECONOMISTS AND THE ENVIRONMENT (1981), at 57–77;
Frances Olson, The Family and the Market: A Study of Ideology and Legal Reform, 96 HARV. L. REV. 1497
(1983), and Margaret Jane Radin, Market Inalienability, 100 HARV. L. REV. 1849 (1987).
debtor–creditor relationship is somehow more destructive of familial feeling than are the indirect and
contingent obligations that arise from a guaranty, then guaranties will help conserve on such psychic
and emotional costs in the same way they conserve on liquidity costs. Given the possibility of default,
there is always a chance that any guaranty will turn into an unconditional debt, with resulting cost to
both fisc and fabric of the family unit. But these costs can be substantially discounted if the
probability of incurring them is low.
D. Why doesn’t the guarantor simply act as an intermediary?
We are not quite done with the analysis. The guarantor’s comparative advantage in informational
monitoring explains why she should bear the residual risk of the debtor’s default; and the creditor’s
comparative advantage in liquidity explains why it should obtain the funds. But neither of these
explanations tells us why the relationship should take the form of a guaranty, for these same
advantages could be achieved through pure intermediation or on–lending. Specifically, C could
obtain funds and lend them to G, who would then turn around and lend to D.79 Under such an
arrangement, G would bear the costs of monitoring D and dealing with any default on his part, since
she would remain liable to C whether or not D managed to repay. There would, moreover, be only
two relationships to manage and monitor as opposed to three. What, then are the advantages of
guaranties as opposed to such intermediation? Or to put the question differently, what are the
advantages of giving C a direct right of recovery against D as well as against G?
It turns out that there are two answers to this question: one applicable in the ordinary situation
where the guarantor is fully able to meet all her financial obligations, and one applicable in the
contingency that she is not. In the former case, the advantage of a guaranty is that the guarantor’s
comparative advantage in monitoring may be incomplete. Recall that monitoring as we have used
the term incorporates a variety of distinct business functions: underwriting, credit investigation,
drafting and executing contracts, auditing, superintending assets, collection, salvage, and the ultimate
bearing of risk following default. It is quite possible for G to have a comparative advantage in some
of these tasks and C to have a comparative advantage in others; and if so, it will pay to have C and
G specialize in different types of monitoring. For instance, consider a loan made for the purposes of
financing a startup enterprise. The borrower’s family members may be better able to assess his overall
79. Or alternatively, we could reverse the timing and have G first lend to D, and then replenish her supply of
liquid funds by borrowing from C. The point is that C could supply the ultimate source of funds without
having to deal directly with D.
honesty and industriousness, keep track of his personal assets, and collect from him in the event of
a business failure. On the other hand, they may know relatively little about the specific line of
business he is contemplating or may lack experience in evaluating whether such startup projects have
what it takes to succeed. For these latter reasons, they may not wish to incur the risk of financing
the project out of their general pool of credit. Once the project has been evaluated and underwritten
by an commercial lender with a background in the relevant industry, however, the residual risk to the
family members of guaranteeing the borrower’s debt is substantially lessened.
One way to achieve such specialization in monitoring, of course, would be through direct
subcontracting of the relevant activities; one party could make the loan, while hiring the other to
conduct underwriting or collection. Indeed, such subcontracts are not uncommon, as the existence
of credit reporting companies like TRW illustrates. But the very fact that a party has a comparative
advantage in some aspect of monitoring makes it difficult for others involved to tell whether he or
she is carrying out that monitoring effectively. For that reason, it is typical to assign hired monitors
some portion of the risk they are hired to monitor; for instance, a collection agency hired to hunt
down delinquent accounts might be paid a commission based on the amounts it actually manages to
extract from recalcitrant debtors. In order actually to make use of C and G’s comparative advantage
in different types of monitoring, accordingly, it is necessary to provide both of them with incentives
optimally to perform their assigned tasks.
A guaranty promotes such dual incentives better than pure intermediation does. Under
intermediation, the risk of D’s default is borne almost entirely by G, and hardly at all by C. C has no
direct connection with D and no specific obligation to monitor him; so long as G retains other assets
sufficient to repay her loan, C is fully protected against loss. Under a guaranty contract, in contrast,
both parties have good incentives to engage in monitoring. Specifically, G’s incentives to monitor
D are provided by the fact that, absent some excuse or defense, she will be liable for his default. C’s
incentives, conversely, are provided by the fact that if it fails to act reasonably or to live up to its
other contractual obligations, G will acquire just such a defense. For under the law of suretyship and
guaranty, C owes special duties toward G — duties it does not owe to other ordinary creditors or
debtors. These are the so–called “suretyship defenses.” Failing to perform duties owed the debtor,
failing to maintain rights against the debtor or against any collateral, granting the debtor unreasonable
extensions or modifications — all operate in the guaranty context to release the guarantor from her
obligation. As §37 of the Restatement provides: “If the obligee acts to increase the secondary
obligor’s risk of loss by increasing its potential cost of performance or decreasing its potential ability
to cause the principal obligor to bear the cost of performance, the secondary obligor is
discharged...”80
The law of suretyship, accordingly, provides C and G with incentives to engage in monitoring in
the same way that the law of negligence provides dual incentives to take precautions against
accidents, or the duty to mitigate damages provides dual incentives to lessen the costs of contract
breach. Under all these regimes, one party — the victim in tort, the promisor in contract, the
guarantor in a surety relationship — is presumptively liable for a loss, and so has optimal incentives
to take care to prevent it. The presumption is reversed, however, if the other party breaches a legal
duty of care. The threat of that reversal gives that other party — the potential tortfeasor, the contract
promisee, the primary creditor C — optimal incentives to take care. The result is what may be called
“double responsibility on the margin.”81
Of course, it is not always efficient to try to provide the parties with such double responsibility
as a legal matter. If there is relatively little that one side can do to influence the underlying risk, or
if the tribunal charged with enforcing the relevant duty of care is not good at determining whether
it has really been breached, then the attempt to provide dual incentives may be more costly than it is
worth. This is why, for instance, strict liability in tort is sometimes more efficient than a negligence
rule.82 Similarly, if one of the parties already has good incentives to take care based on extralegal
considerations such as reputation, it may be better for that party to disclaim liability for negligence
or similar breaches of duty, reserving legal sanctions for the purpose of motivating the party not
80. RESTATEMENT (THIRD) OF SURETYSHIP AND GUARANTY, §37 (1996) [“Impairment of Suretyship Status”].
For the content of suretyship defenses more specifically, see id., §§38–49.
81. See generally Robert Cooter, Unity in Tort, Contract, and Property: The Model of Precaution, 73 CALIF. L.
REV. 1 (1985).
82. See Steven Shavell, Strict Liability Versus Negligence, 9 J. LEGAL STUD. 1 (1980) [formal comparison of
precaution incentives under competing rules]; Guido Calabresi and Jon Hirschoff, Toward a Test for Strict
Liability in Torts, 81 YALE L.J. 1055 (1972) [informal comparison of incentives]. Similar considerations of
relative ability to reduce risk and institutional capacity also determine which party should be made
presumptively responsible for a loss under a negligence–type rule, and which party’s breach of duty should
operate to reverse that initial presumption. For instance, if courts are extremely good at detecting and
punishing unreasonable behavior on the part of injurers, negligence is the most efficient rule of tort liability.
If conversely they are better able to evaluate victims’ behavior, it is better to have a rule of strict liability with a
defense of contributory negligence. Id. In the guaranty context, analogous factors determine which of C and
G should take the role of creditor, and which the role of guarantor.
subject to such extralegal incentives.83 For all these reasons, then, it may be desirable to cast a credit
transaction in the form of intermediation rather than a guaranty. Such factors can also help explain
why it might be efficient for guarantors to waive their rights to assert the standard suretyship
defenses; indeed, such explicit waivers are routine in many important commercial settings.84
The foregoing analysis applies to the extent that G stands ready to perform her financial
obligations in full. It is always possible, however, for G herself to turn out to be uncollectible in the
event of D’s default. While debtors and their guarantors hold entitlements to distinct groups of assets
— else there would be no point in incurring the transaction costs of a guaranty in the first place —
the values of these respective assets will in general be correlated. The fortunes of families tend to rise
and fall together; so do the financial situations of affiliated companies, of banks and their depositors,
and of suppliers and their customers. Indeed, even if the risks of D’s and G’s insolvency are totally
independent, there will still exist a possibility that G will be unable or unwilling to make good on her
promises to C.85 The same is true, of course, if the transaction takes the form of pure intermediation.
In such event, however, guaranties and intermediation have different consequences. Under
intermediation, C will have no contract or property claim allowing it to reach any of G’s assets in
particular. In order to collect from D directly in such an instance, rather, C would first have to win
a lawsuit against G for breach of contract, and then obtain and enforce a judgment lien against G’s
interest in D’s loan. Such remedies, however, can be costly and, more importantly, in the time it
takes under state collection law to pursue them, any value remaining in the loan could be dissipated.
Furthermore, if G actually turns out to be insolvent, C will be forced to participate in bankruptcy
proceedings and will have to share any proceeds arising from D’s debt (and from G’s other remaining
83. See, e.g., Benjamin Klein, Transaction Cost Determinants of ‘Unfair’ Contractual Arrangements, 70 AM.
ECON. ASS’N PAPERS & PROCEEDINGS 356 (1980) [defending unrestricted right of franchisors to terminate
franchise contracts without proof of cause, on grounds that their tendencies toward opportunism are restrained
by reputation while franchisees’ are not].
84. See sources cited in note 40, supra. Admittedly, such waivers could also result from guarantor unsophistication
or lender overreaching; on this possibility see note 46, supra, and accompanying text.
85. This risk is nontrivial even when dealing with large commercial guarantors and issuers of standby letters of
credit, as the 1982 failure of Penn Square Bank illustrated. See Michael A. Goldberg and Peter R. Lloyd-
Davies, Standby Letters of Credit: Are Banks Overextending Themselves?, 16 J. BANK RESEARCH 28 (1985).
assets) with all of G’s other creditors on a pro rata basis.86 Under a guaranty, in contrast, G gets an
expedited collection procedure and an effective priority in D’s debt over G’s other creditors.87
These last observations suggest a second and distinct advantage of guaranties. Even if G is better
at all aspects of monitoring than C, there remains a chance that G will be uncollectible while D is not.
In this event, C will value the option of obtaining repayment directly from D; and under a guaranty,
it has that option. Of course, giving such an option to C is not costless; to the extent that C is able
to satisfy itself out of the proceeds of D’s debt to G, there will be fewer assets left for G’s other
creditors. Whether it is efficient to give C a priority in those proceeds, however, depends on whether
that priority is more valuable in C’s hands than in the hands of G’s other creditors — or equivalently,
whether C gains more from the priority than the other creditors lose.
But there already exists a well–established account of how one should allocate priority among
creditors in the event of insolvency; it can be found in the economic literature on secured credit. This
account begins with the observation that secured credit has higher direct transaction costs than
unsecured credit: it is necessary to write a special contract, file a public notice, restrict the use of
collateral, and so on. Security is only worthwhile, on this account, if its other advantages justify these
costs. This can be the case only if the secured lender places a higher value on a specific interest in
the collateral than do the debtor’s other creditors; else the cost the debtor will pay when seeking other
financing after having encumbered her assets will more than outweigh the benefits she receives in
exchange for granting a security interest.88 The secured lender’s higher value in specific collateral can
arise from various factors: from superior information regarding its market or use–value, from a
comparative advantage in salvage, from superior ability to monitor the collateral’s whereabouts or
to prevent it from being diverted to other uses — or from a relatively inferior ability to monitor,
evaluate or make use of the debtor’s other assets. If any of these factors are present, value is created
86. See 11 U.S.C. §726 [dealing with distribution of property in Chapter 7 liquidations]; 11 U.S.C. §§1122–1129
[dealing with requirements of Chapter 11 reorganization plans].
87. Although both these advantages may be limited in degree by federal bankruptcy law. See, e.g., Trimec, Inc. v.
Zale Corporation, 150 B.R. 685 (N.D. Ill. 1993) [applying automatic stay provision of the Bankruptcy Code,
11 U.S.C. §362, to defer a creditor’s attempt to collect from a nonbankrupt debtor, where such debtor shared
substantial business identity with a bankrupt guarantor].
88. See Schwartz, A Theory of Loan Priorities, supra, note 3.

when the secured lender takes priority in the collateralized asset, leaving other, unsecured creditors
to take their compensation through other property or through an increased price or risk premium.89
The choice between intermediation and guaranties is precisely analogous to the choice between
unsecured and secured lending. Guaranties have higher direct transaction costs than intermediation.
While no public notice is required to create a guaranty, one does need to write a special contract,
keep track of rights and duties of three parties, and restrict the manner in which the guarantor and
creditor deal with the debtor. Guaranties are only worthwhile compared to intermediation,
accordingly, if C places a sufficiently higher value on a having a direct claim against D than do G’s
other creditors; else the costs that G incurs in her dealings with these other creditors will outweigh
any benefits she gets when giving or selling the guaranty. This condition will be satisfied if C has
superior information than those other creditors regarding the value of D’s obligation to G, if C has
superior ability to supervise D’s behavior, or C has superior ability to collect from D — in short, if
G has a sufficient comparative advantage in monitoring D compared to G’s other creditors.
The condition seems to be met in most cases in which guaranties are used. Because C has had
some prior dealings with D, has investigated D, enjoys some special leverage over D, or expects to
do any of these things in the future, it makes sense for it to look specifically to D out of all of G’s
assets, rather than to G’s assets in general. As a simple illustration, consider the position of an
automobile dealer who lends to a college student on a loan co–signed by the student’s parent, where
the parent’s other main liability arises out of the mortgage on the family house. The auto dealer will
place a much higher value on the right to proceed against the student than will the parent’s mortgage
lender, for the latter does not know anything about the student and may not even be accustomed to
dealing with cars. Accordingly, it loses the mortgagee little to allow the auto dealer priority in
whatever value may lie in the student’s promise to pay.
Similarly, consider the case of a recourse sale of consumer accounts. As we saw in the previous
subsection, recourse financing corresponds to a three–corner guaranty; the consumers play the role
of D, the account factor the role of C, and the retailer, who bears the ultimate risk of uncollectible
89. See generally Schwartz, A Theory of Loan Priorities, supra, note 3; Triantis, supra, note 3; Scott, supra, note
3; Buckley, supra, note 4. In addition, as Ronald Mann has argued, secured credit may also, by hindering the
flexibility of the debtor's business, have the advantage of providing a credible commitment against excessive
future borrowing. See Mann, supra note 68; Ronald Mann, Explaining the Pattern of Secured Credit, 110
HARV. L. REV. 625 (1997) [arguing that this last consideration provides the best explanation of the empirical
evidence relating to the use of security.]
accounts, the role of G. Nonrecourse financing, where the factor both provides funds and bears the
residual risk of default, corresponds to a direct loan from C, the factor, to D, the consumers. The
direct loan from G to D was mirrored in the possibility of not factoring the accounts at all. What
retail financing transaction, then, corresponds to pure intermediation or on–lending?
Recall that with on–lending, the creditor C provides funds relying only on G’s commitment to
repay; G then uses those funds to finance D’s borrowing. The analogous arrangement here is for the
factor to lend to the retailer on the strength of her general credit, without taking any specific interest
in the retailer’s contract rights against her customers, and for the retailer to sell on credit to
consumers. Such an arrangement makes little sense when dealing with a professional factor,
however, because it sacrifices the factor’s valuable expertise. The reason that a retailer can get more
from a factor for her accounts than she can from the undifferentiated run of unsecured creditors is
that the factor specializes in account financing and is expert in assessing and monitoring their value.
Other lenders such as finance companies and trade creditors, who know relatively little about
accounts or their collection, will only lend against them at an extreme discount. Accordingly, it
makes sense for the factor to have a direct right of action (and perhaps a security interest as well) in
the accounts, rather than sharing them with these other lenders.
The harder cases for guaranties are those in which C and D have had no prior dealings, as when
the guaranty relationship is established through an assignment of contract rights or a delegation of
contract duties. But even in such cases, it may be better to have C look to both G and D for
performance, rather than just to G alone. Consider as an illustration a sublease on an apartment,
where the landlord has no role in choosing the subtenant. In this situation, the parties could arrange
to have the subtenant primarily liable to the landlord for the rent and the tenant secondarily liable —
a guaranty arrangement — or they could have the subtenant liable only to the tenant, who would in
turn be unconditionally liable to the landlord — a form of intermediation. Most likely, however, the
parties as a group will be better off having the landlord look for payment to both the original tenant
and her subtenant. The tenant, who has moved to another location and taken her personal
possessions with her, may well prove uncollectible beyond the amount of her security deposit. The
value of any payments made by the subtenant, conversely, will likely depend on how well the landlord
keeps up its corresponding obligations under the primary lease. Similarly, the landlord, through its
management of the building and adjoining realty, has information regarding the subtenant’s activities
and whereabouts, and control over any misbehavior on his part that the original tenant’s other
creditors lack. All this causes the landlord to place a higher value on the subtenant’s payments than
would these other creditors, making it efficient for it to have priority over them in the rent.
Guaranties are more efficient than pure intermediation, in short, when C has some special interest
or comparative advantage in monitoring D — either relative to G, or to G’s other creditors. The
former comparison will be relevant in the event that G is solvent, and the latter comparison will be
relevant in the event that she is not. Of course, since there is some chance of either of these situations
arising ex ante, both will bear on transactional planning in proportion to their respective probabilities.
Intermediation, in contrast, makes sense when the ultimate creditor has no comparative advantage
in monitoring the ultimate borrower. As a stereotypical example, consider commercial banking as
it has traditionally been practiced: the bank obtains liquid funds from various individual depositors
and lends them out to various business and consumer borrowers. Here the plainly efficient
arrangement is for the bank to act as a pure intermediary. Since the depositors are in no position to
evaluate or supervise the bank’s lending decisions, there is no reason for them to take priority in or
to have a direct claim against any particular loan or group of loans. The transaction costs of doing
so — including, most importantly, the substantial inconvenience to the depositors of having to
concern themselves with the bank’s lending portfolio, and the corresponding inconvenience to
borrowers of having to deal with inquiries from a multitude of small, scattered creditors — would
simply be wasted.90
IV. Applications, public and private
To summarize the argument thus far: the previous sections of this paper have set out a theoretical
framework that can serve to unify our understanding of guaranty transactions across a wide variety
of legal and policy contexts. This current section seeks to show how that framework can also be used
to improve the quality of transactional planning. It focuses on two extended examples: first, we will
consider how our economic framework bears on the optimal design of government–provided credit
assistance. Second, we will consider what conditions, if any, should be placed on the liability of
commercial issuers of standby letters of credit.
90. For a formal economic model reaching similar conclusions, but which focuses on the bank’s ability to diversify
risk as an explanation of its incentives to monitor effectively, see Douglas W. Diamond, Financial
Intermediation and Delegated Monitoring, 51 REV. ECON. STUD. 393 (1984).
In addition to providing one problem drawn from the public policy arena and one from private
commerce, the examples are also intended to illustrate how our overall theory can be applied at
various stages of planning and levels of detail. The discussion of Federal credit policy, accordingly,
stands for the broader problem of choosing basic transactional form; that is, should such credit
assistance be provided though guaranties, through direct government lending, or through a program
of outright grants and subsidies? To address this broader problem, we will need to apply the analysis
of the previous section at a general level: asking who is the least–cost monitor, whose cost of liquidity
is lowest, and who places the highest value on a claim against the debtor. The example of standby
letters of credit, conversely, stands for the narrower problem of choosing which individual incidents
of suretyship should be included or excluded in a particular transaction, once the decision has been
made to use the form of a guaranty. To address this latter problem, accordingly, we will need to
focus on more detailed considerations of comparative advantage: what are the specific risks the
transaction is meant to address, what are the particular types of monitoring to be undertaken, and
which of the parties is best placed to bear each of these individual obligations?
One caveat before beginning, however: for both these examples, the discussion that follows
should be understood as a mere sketch of the sort of inquiry that would be needed to resolve such
questions in the context of actual decisionmaking. For a policymaker to authorize the commitment
of public resources, or for a private attorney to induce the trust and reliance of clients, he or she
ought to have substantially more detail factual support than can reasonably be developed in the space
available to us here. The point of this exercise, then, is not to reach definitive conclusions, but rather
to set out a method by which such conclusions might ultimately be reached.
A. The government as guarantor
Many government policies and expenditure programs are designed with the purpose of assisting
individual borrowers; beneficiaries of such programs include farmers, small businesses, students,
home buyers, U.S. exporters and their foreign customers, banks, municipalities, and independent
sovereign governments. The motives for providing such assistance are diverse. They range from
redistribution (e.g., farm credit, urban enterprise zones) to subsidizing purchases deemed to have
some significant public good aspect (e.g., home ownership, foreign aid) to correcting some externality
or informational failure in the relevant credit market (e.g., deposit insurance).91 Some policies, such
91. For a rigorous exposition of the information failure rationale, see N. Gregory Mankiw, The Allocation of
Credit and Financial Collapse, 101 Q. J. ECON. 455 (1986)
as support for educational borrowing and the much–debated bailout of the Chrysler Corporation in
the late 1970's, can be defended (or attacked) on all these grounds.
As we have seen, however, there are various means through which credit assistance can be
provided. Most straightforwardly, the state could directly subsidize interest payments, either by
outright cash transfers to borrowers or through some sort of tax expenditure. The federal income
tax deduction for residential mortgage interest provides the classic example of this first strategy.92
Alternatively, the state could itself lend funds to the intended beneficiaries at more generous terms
than private creditors would be willing to offer; the National Direct Student Loan program is the
model here. And perhaps most indirectly, the state could let beneficiaries borrow on the open market
but assume a portion of their risk of default by offering a guaranty, as with the Chrysler and Mexico
bailouts. Each of these approaches would lower the beneficiaries’ effective price of credit; from the
government’s perspective, each would entail some expected cost. The payments that would actually
need to be made ex post under these three schemes, however, could differ substantially in timing and
variance.
In the areas of housing, higher education, foreign aid, and assistance to small business, all three
of these approaches have been tried over the years. More recently, however, there has been a distinct
trend away from implementing such programs through direct expenditures, and toward implementing
them through loans or guaranties. Obviously, many of the causes of this trend have little to do with
efficiency. Casting a credit subsidy in the form of a guaranty or a loan has many practical political
advantages; it can reduce the subsidy’s visibility to potential opponents or competitors, vest the
jurisdiction of the program in one agency or Congressional committee rather than another, or assist
its proponents in logrolling a majority coalition by manufacturing an opportunity to share the wealth
with other organized political constituencies such as institutional lenders. Furthermore, the current
political context may lend such considerations special weight. Popular resistance to taxes and to new
government spending is probably the most salient feature of contemporary American politics; it
explains a wide variety of recent events such as the growth of deficit spending, the move to increased
tax expenditures, and the various attempts to limit the fiscal power of government through
constitutional and quasi–constitutional procedures such as the balanced–budget amendment, the
Gramm–Rudman–Hollings Act, and the line–item veto. In such an environment, there is always
92. Alternatively, such a tax expenditure could take the form of a tax credit (refundable or otherwise), or of
deferral or exclusion of income ultimately produced by the underlying project for which the loan is extended.
strong pressure to structure new initiatives, and to restructure new ones, in a form that minimizes the
constraints posed by formal and informal budgetary procedures. For a variety of institutional reasons
— some statutory, some administrative, some simply political — loans and guaranties have often been
easier to enact and implement than direct expenditures.93
While it would be unrealistic to wish such considerations entirely away, it is still worth
investigating which of the possible methods by which the government subsidizes credit is most
cost–effective. At least for certain categories of credit, the choice among these methods remains a
live political issue. In 1993, for instance, Congress overhauled the federal student loan program,
significantly substituting direct lending for guaranties.94 While this policy change was largely justified
by its proponents on grounds of administrative efficiency, critics of the change questioned at the time
whether these efficiencies were likely to be realized in practice;95 and economic studies that have since
appeared suggest that the net outcome was in fact to increase the costs of the program.96 A fuller
analysis that made use of our conceptual framework could have improved the quality of the debate
by helping to clarify the specific costs and benefits of these alternative transactional forms.
Moreover, recent policy initiatives toward “reinventing government” and rationalizing the
budgetary process hold out some hope that the design of future credit assistance programs can be
evaluated on the basis of their substantive economic merits. Under the Federal Credit Reform Act
of 1990, for instance, guaranties and other contingent government obligations are no longer
93. See, e.g., Stith, supra, note 54, Kate Stith, Rewriting the Fiscal Constitution: The Case of Gramm-Rudman-
Hollings, 76 CALIF. L. REV. 595, 605–621 (1988); H. Leonard, CHECKS UNBALANCED (1986) (discussing
political appeal of various forms of indirect and off–budget government spending, including loans and
guaranties). Note that incentives may be shifting in this regard; with rise of GOP Congress since 1995, strong
tendency to cast such programs as tax expenditures that can be characterized in election campaigns as tax cuts.
94. Omnibus Budget Reconciliation Act of 1993, Pub. L. 103–66, §13131(b), 107 Stat. 312, 342–43 (1993). The
changeover was phased in over a five–year period, with the result that in fiscal 1998 60% of federal student
loans are to be made on a direct rather than a guaranteed basis. See Brody, supra, note 7, at 507.
95. Id., at 507–509.
96. See Zimmerman and Miles, supra, note 7; Miles and Zimmerman, supra, note 7 (finding that budget savings
from direct lending were overestimated because credit risks absorbed by the federal government were ignored,
budget scorekeeping rules pushed recognition of some administrative costs beyond the five-year budget
window, and potential increases in administrative costs and government borrowing costs were ignored).
automatically treated as off–budget.97 Instead, the estimated net present value of any new loan or
guaranty, taking into account factors such as the interest rate, repayment period, and risk of default,
or of any action that alters the terms of existing loans or guaranties, must be reflected in the
President’s budget and go through the Congressional appropriations process.98 Admittedly, the Act
does not apply to all federal guaranties, specifically exempting student loans, veterans' home loans,
and other entitlement programs from its coverage.99 Other loopholes and accounting devices that
privilege guaranties over direct expenditures still remain.100 Nonetheless, such reforms suggest that
there is some political support for closing such loopholes and eliminating such privileges. As the
sophistication of financial regulators continues to improve, and as the private sector also develops
more accurate methods of accounting for contingent liabilities,101 candid evaluations of the
comparative transaction costs of alternative government programs should increasingly find an
audience.
Accordingly, consider how such programs could best be structured, focusing on the specific
examples of higher education, housing purchases, and foreign aid. As the previous section of this
paper demonstrated, the answer to this question turns on two key factors: the comparative costs of
liquidity and of monitoring debtors. More specifically, if the government has lower liquidity and
lower monitoring costs than do private lenders, then other things being equal direct government loans
are the most efficient means of assistance. If private lenders have lower liquidity and monitoring
costs, conversely, it is most efficient for the government directly to subsidize interest payments on
private loans. Government guaranties are best if the government has a comparative advantage in
monitoring while private lenders have an advantage in liquidity. And if the government has lower
97. Federal Credit Reform Act of 1990, Pub. L. No. 101-508, 104 Stat. 1388, 1388-610 (1990) [codified as
amended at 2 U.S.C. 661 (1997)].
98. Id. §661(c).
99. Id. §661c(c)(1).
100. See, e.g., Humphrey, supra, note 21, at 196–200 [describing how Clinton administration succeeded in putting
together multilateral aid package of over $51 billion, including $20 billion in currency swaps and loan
guaranties from the U.S. Treasury and Exchange Stabilization Fund, despite Congressional and popular
opposition]; Zimmerman and Miles, supra, note 7, at 780 [describing how budget scorekeeping rules put in
place under the FCRA resulted in understating the administrative costs of direct loans, most of which are
incurred in years beyond the five–year budget window].
101. See discussions at notes 50, 51, and 56, supra.

liquidity costs but higher monitoring costs, then the government ought to make the loans, but have
the private sector guarantee them and bear the risk of default.
Let us then assess these two factors in turn. First, does the government have any comparative
advantage in liquidity? The answer is unclear. The government can borrow at a lower interest rate
than private borrowers can, to be sure, because its obligations are perceived by the debt market as
virtually risk–free.102 In contrast, even a loan to an established bank or blue–chip corporation entails
some chance of default.103 This risk differential, however, does not reflect a true difference in cost
from the social point of view. The reason that U.S. government bonds are deemed less risky than
private bonds is that the government has a clear and credible option that private borrowers who get
into financial trouble lack — namely, it has the power to raise funds by levying taxes or printing
money. But these sources of revenue are not costless. When the government raises taxes, it imposes
liquidity costs on the citizens whose tax liabilities are increased, and engenders further efficiency costs
in the form of deadweight loss or excess burden.104 Similarly, when the government prints money,
it imposes a real tax on owners of cash and other money balances whose holdings are consequently
devalued, and redistributes wealth from creditors whose contracts are denominated in nominal terms
to their debtors; both of these consequences create real inefficiencies and the former reduces the
liquidity of cashholders. In addition, both levying taxes and printing money can have serious and
102. Save the risk of inflation, of course. This risk, however, can be largely eliminated by indexing the schedule
of payments for inflation or by contracting for a variable interest rate. The Treasury has recently announced a
plan to start selling just such variable–rate obligations. [Cite contemporaneous news accounts.] Whatever the
fate of this plan, however, (it met with considerable skepticism from relevant Congressional and regulatory
actors at the time) roughly equivalent hedges against inflation are widely available on private financial
markets.
103. While it is theoretically possible that the federal government could someday choose to default on its general
obligations, the market rightly judges the chances of this contingency, given political realities, to be trivial.
(Recall how during the 1995–1996 shutdown of the federal government, both Congress and officials in the
executive branch took great pains to assure financial markets that funds would be available without
interruption to pay off government bonds as they came due. [Cite from contemporary press accounts.] The risk
of gradual repudiation of such debts through a policy of deliberate inflation is a more serious one; still, the
vesting of discretion over the money supply in the Federal Reserve, a legally and politically independent
agency, substantially reduces (while not eliminating) the risk that monetary policy will be used for such
opportunistic ends.
104. The deadweight loss or excess burden of a tax is the efficiency loss caused by taxpayers substituting away
from the taxed activity in favor of another, less high taxed activity — for instance, working fewer hours under
a wage tax, saving and investing less under a tax on interest income. See generally Harvey Rosen, PUBLIC
FINANCE ( 3d ed. 1992), chap. 4 (“Tools of Normative Analysis”).
adverse macroeconomic effects, crowding out investment and worsening the tradeoff between
unemployment and inflation.105 Properly calculated, then, the government’s liquidity costs include
all of these potential consequences of increased borrowing, discounted for their respective
probabilities and expected time of occurrence.
As a theoretical matter, if the government always acted with perfect information and rationality,
its expected cost of borrowing would have to be lower than that of private sector. The reason is that
the government’s option to raise funds via taxes or the printing press — like all options — has a
positive expected value. If the option turns out to be less expensive than other sources of liquidity,
it can be exercised, while if it turns out to be more expensive, it can be freely ignored. For a fully
informed rational actor, it follows that having an option is always better than not having one. But of
course, the government does not always act perfectly when making fiscal or monetary policy.
Because of imperfect information, agency problems, the absence of any property rights in the
surpluses that could be created by efficient policies, and other costs of administration, the option of
tax–based or money–based finance does not necessarily translate into reduced borrowing costs.
Coase’s critique of the theory of market failure is apposite here; it is misleading to focus on
deficiencies in private credit markets while ignoring the transaction costs of the government’s
financial activities.106
As a first approximation, therefore, it is reasonable to treat the government’s cost of liquidity, all
things considered, as roughly the same as that of private lenders. In any event, I will make this
assumption in what follows, while recognizing that the question is an empirical one and there is room
to disagree. Indeed, one’s views regarding current macroeconomic policy can serve as a rough proxy
for an answer to the comparative liquidity question. If one believes that the government’s marginal
costs of liquidity are really less than the private sector’s, then it follows that the government ought
to increase its overall lending to the private sector compared to current levels, or, what is the same
thing, reduce its overall borrowing. If one thinks, however, that the current level of government debt
is approximately optimal, however, than it is hard to see where any liquidity advantage from shifting
between government and private credit would come from.
105. Cite any macro textbook on these propositions, especially one by Barro or one of the other new classicists.
106. See generally Ronald Coase, The Problem of Social Cost, 3 J. LAW & ECON. 1 (1960).
Turn then to the other half of our inquiry: is there reason to think that the state can engage in
monitoring more cheaply than private lenders? In the areas of small business lending, student loans,
and housing, such a general claim seems problematic. Consider first the component of underwriting
and risk assessment; the available evidence suggests that, if anything, government credit standards
are weaker than those of the private sector. The government’s debt pool is riskier than that of private
lenders’; and it tends to suffer from disproportionate default. Some of this risk differential, of course,
stems from the adverse selection that comes with being a lender of last resort. But surveys of the
government’s actual credit practices tend to show that, if anything, it makes lending decisions based
on less information and less sophisticated methods of evaluation than do private lenders.107 Indeed,
it is hard to see how it could be otherwise. Many of the government’s credit assistance programs
have been created as entitlement programs, so that both legal authority and the political atmosphere
encourage leniency in applying program criteria. There is no obvious capacity the government has
in this regard that private auditors lack; and the latter have the advantage of greater experience, more
up–to–date information technology, and the incentives for honing their techniques that come from
the spur of the profit motive in a competitive market.
With regard to supervision of the debtor during the life of the loan and collection following
default, the case for governmental comparative advantage is only marginally more compelling.
Possibly the federal government can more cheaply trace debtors’ or their assets across state lines —
an argument that has frequently been put forward in the context of unpaid child support — but the
actual opportunity costs of federal investigative and policing resources would need to be more fully
accounted for in considering such a claim. The government’s collection costs could also be lower
than the private sector, especially if such costs were combined with or offset against other public
collection or payment activities. Recent proposals to use the Internal Revenue Service to collect
unpaid student loans illustrate this last possibility; similar economies of scope might also be achievable
through the welfare or Social Security systems. In this regard, however, it is worth noting that the
IRS has fought all such proposals to draw on their enforcement resources tooth and nail; and other
agencies such as the Social Security Administration can be expected to offer similar resistance.
107. See, e.g., Zimmerman and Miles, supra note 7, at __; U.S. General Accounting Office, High–Risk Series:
Guaranteed Student Loans, GAO/HR–93–2 (1992), at 30–31 (“... the inventory of known problems in the
[Education] Department’s administration of guaranteed student loans raises questions about its ability to
adequately manage a direct lending program.”)
In the end, accordingly, the case for public provision of such functions must be made on a more
detailed accounting of the relative administrative costs in private and public settings. For some types
of collection, such as payroll deductions for loans made to government employees or persons serving
in the military, the government may have a real comparative advantage. More generally, however,
proponents of bringing credit monitoring services within the government sector should probably be
assigned the burden of proof — especially in light of the unwarranted claims of administrative
efficiency that they recently made in the area of direct student loans.
There are two specific instances, however, in which the government may have a real comparative
advantage in monitoring over private lenders. One of these arises in contexts where the justification
for government subsidy stems from racial, ethnic, or gender discrimination; this justification has
commonly been offered in markets for home mortgages and, to a lesser extent, small business loans.
If private lenders discriminate on such grounds, whether out of animus, stereotype, or cultural
distance from borrowers, then government agents less subject to such cognitive limitations may do
a better job of gauging default risk and auditing ongoing loans.108 In such cases, accordingly, loan
guaranties may be more cost–effective than straight interest subsidies, and direct loans more effective
than guaranties, in providing discriminated–against borrowers with credit.
The argument here for direct loans or guaranties, however, is a fairly narrow one. Discrimination
in lending markets can occur for a number of reasons: bias in evaluating credit risk, discomfort on the
part of lending officers when dealing with particular classes of applicants, lenders’ inexperience in
evaluating property values in certain neighborhoods or certain categories of assets such as “sweat
equity,” leading them to add in an extra markup reflecting the associated risk premium from their
standpoint. Or it could simply result from the fact that the discriminated–against applicants have
lower wealth or propose to invest in neighborhoods that are economically more marginal. For the
government to have a comparative advantage in monitoring, two conditions must be satisfied: first,
the cause of discrimination must be noisy information rather than a simple reluctance to lend; and
second, the government’s assessments of risk must have lower variance than those of private lenders,
not just lower mean. Unless the government is actually more accurate than private lenders at
108. See, e.g., Charles W. Calomiris, Charles M. Kahn, and Stanley D. Longhofer, Housing–Finance Intervention
and Private Incentives: Helping Minorities and the Poor, 26 J. MONEY, CREDIT & BANKING 634 (1994)
[presenting and analyzing different several economic models implying such a result].
distinguishing between high–risk and low–risk applicants, it would do better in efficiency terms simply
to provide a general interest subsidy for applicants in the relevant group.109
Second, the government may have under consideration other programs that will affect the ability
of borrowers to repay. For instance, during the period in the late 1970's in which the Chrysler
guaranties were extended, the federal government had several opportunities to revisit its automobile
emissions and fuel economy standards, as well its program of voluntary import restraints that
effectively placed a quota on the number of Japanese cars sold in the United States.110 All these
policies had the potential for a major effect on Chrysler’s profits; and the fuel economy standards in
particular arguably played a key role in helping Chrysler to reverse its fortunes by competing more
effectively with the other Big Three auto companies. When government has the power to influence
the risk of a borrower’s default in this fashion, private lenders face a relatively high cost in monitoring
and bearing that risk. Just as a guaranty from a corporate manager or other insider gives the insider
to take stronger precautions to ensure the corporation’s ability to repay, so would a government
guaranty provide the government with more credible incentives to implement risk–reducing policies
and refrain from risk–increasing ones.
Whether such incentives are actually welfare–enhancing on balance, however, is unclear; the
government’s potential liability on its guaranties may lead it to take an excessively short–term or
narrow financial perspective, distorting its decisions on other policies. The relationship between US
air pollution standards and Chrysler’s financial health in the 1970's and 80's illustrates the potential
for such a conflict of interest. Furthermore, this argument threatens to swallow up the entire private
market for credit and to force the government to stand as potential guarantor for all debt contracts
in the economy, for when the overall effects of fiscal and monetary policy on interest rates and
unemployment are taken into account, there is virtually no loan whose risk is not influenced by
109. Indeed, here is an instance where economic efficiency and procedural norms of equal treatment may be in
tension. If the government has reason to believe that women or racial minorities are unfairly discriminated
against by lending officers on average, for instance, but is no better than these same lending officers at
distinguishing high–risk from low–risk applicants within demographic categories, the more efficient policy
would be to subsidize all loans made to women and minorities. Such a policy, however, would undoubtedly be
attacked on grounds of “reverse discrimination.” A more facially neutral policy of guaranteeing all applicants
regardless of race or gender would, ironically, be less effective and more costly.
110. See Robert B. Reich, Bailout: A Comparative Study in Law and Industrial Structure,2 YALE J. ON REG. 163,
181 (1985) (recounting background to and assessing Chrysler loan guaranties).
government action. Absent some specially significant connection between government actions and
the risk of default, accordingly, it is probably best on balance to ignore this factor.
Additionally and finally, while we have focused on economic efficiency in the discussion thus far,
even if the government has no particular advantage in monitoring or liquidity, a program of guaranties
may be more desirable than direct loans or general interest subsidies on distributive grounds.
Specifically, guaranties may in some cases have the advantage of better targeting the benefits of
government credit assistance to those borrowers who are deemed most needy or most deserving.111
The benefits of interest subsidies are enjoyed by all borrowers regardless of their risk of default.
Guaranties and direct low–interest loans, in contrast, primarily benefit those borrowers who face high
costs on the private credit market. For the same reason, guaranties and direct loans may be better
at directing benefits to marginal borrowers who could not otherwise obtain credit, while interest
subsidies are paid to marginal and inframarginal borrowers (i.e., those who would be able to obtain
credit even without a subsidy) alike. This last advantage is unimportant if the government has a full
array of alternative redistributive tools at its disposal, or if it is feasible to restrict interest subsidies
to those who could or would not otherwise borrow. In a second–best world where efficiency and
distribution must be traded off for one another, however, it cannot be ignored.112
Apart from these last few considerations, however, the efficiency case for either government
direct loans or government loan guaranties seems weak. If the state has no comparative advantage
in either monitoring or liquidity, it should get out of the lending business. This does not mean that
the government should not support borrowers; rather, it should do so through direct subsidies such
as tax deductions and credits. Indeed, carefully designing such deductions and credits could be a way
of more precisely targeting them to those most in need.113 I suspect, accordingly, that the main reason
111. Thanks to Scott Masten for suggesting this point.
112. Compare Calomiris et al., supra, note 108, who show that if applying for loans is expensive, rationing
subsidies to a random subset of applicants also provides a way to direct assistance to the most needy, by giving
those who could obtain credit on their own an incentive to self–select out of the program. They observe that
delay in processing benefits, furthermore, is a common way that government benefits programs encourage
such self–selection.
113. The personal income tax provides a particularly flexible set of tools through which to do this; credits or
deductions could be limited to borrowers below a certain level of income, could be phased out as the
borrower’s income rises, or could be deferred over a period of years. See, e.g., Brody, supra, note 7
(advocating a system of student loans, administered through the tax system, in which the obligation to repay
would be income–contingent).
for the popularity of direct loans and loan-guarantees stems is not economic but political; more
precisely, the statutory and procedural constraints of the budgetary process described above make
guaranties the cheapest way to assist borrowers through the political process.
It is worth discussing whether this ultimate result could be justified on second–best efficiency
grounds. After all, the constraints of the political process are real constraints; and an economist
interested in maximizing the sum of benefits over costs should take them into account. Similarly, it
might be argued that choosing an inefficient transactional form can be justified on grounds of
paternalism. If credit assistance is sufficiently worthwhile, on this view, an inefficient program of
assistance is better than no program at all; and if concealing the assistance or disguising its costs by
casting it in the form of a guaranty is the only way to get it enacted, so be it. Something like this
seems to have been the rationale of the Clinton administration in fashioning the recent Mexican
bailout, and perhaps also was the Bush administration’s justification for proposing Israeli loan
guaranties some years earlier. Indeed, we might colloquially call this last argument the “Lend–Lease”
argument, in honor of Franklin D. Roosevelt’s famous pre–World War II policy of assisting Britain
and France against Nazi Germany before the American public was ready to support such assistance,
and disguising the extent of the assistance though the use of a legal form.
While misrepresenting the costs and benefits of public programs might be justifiable on pragmatic
grounds in exigent circumstances such as FDR faced, however, paternalism of this sort cannot form
a basis for government policy in general. Since this paper aims to provide a general method for policy
analysis and planning, it must therefore put such considerations aside. It is worth noting, however,
that the utilitarian argument for deception by political elites is particularly weak when used in the
service of policies (such as the student loan program) that command the support of a broad–based
middle–class constituency. Instead, this argument has traditionally been asserted most frequently in
the setting of foreign policy setting, on the grounds that the legislature — and popular opinion as
well — structurally underappreciates the value of expenditures on foreign assistance.114
The foreign policy setting, however, actually offers a more plausible case for the use of loan
guaranties on efficiency grounds. While the argument that the government has a comparative
advantage in liquidity is no stronger here than anywhere else, its advantage in monitoring loans made
114. Though of course there is a similarly long tradition in American politics of celebrating the vox populi as a
bulwark against the danger of foreign entanglements. With regard to these competing traditions, take your
pick.
to foreign allies may be quite real. For one thing, U.S. officials have superior access to at least some
of the political and economic information relevant to the risk of such countries’ default, an advantage
partly based on classified materials, intelligence operations, and confidential diplomatic contacts.
While not all this information is available to the particular actors who are charged with making
lending decisions, those actors still have better access to it than do private lenders. Second and more
significantly, the U.S. government has more leverage than private lenders do in the event of default.
A disappointed private creditor is for most purposes limited to calling in outstanding loans, refusing
future extensions of credit, and publicizing the debtor’s default to other lenders. The U.S. qua lender
can do all of these things as well. In addition, however, because of the many dimensions in which
nations interact in the international arena, it has at its disposal a much wider array of foreign policy
tools to be used as carrots and sticks — diplomatic concessions, reductions in informal trade barriers,
state visits and other gestures of political theatre, the exercise of voting rights in multilateral
organizations, and so on. As a result, the U.S. will generally find it easier and cheaper to collect
following a default — or to turn the default to its advantage by trading debt forgiveness for other
foreign policy concessions it should subsequently find valuable. These comparative advantages in
collection and salvage, accordingly, make it likely that the government will be the more effective
residual creditor.
Furthermore, as the theoretical framework of the previous section suggests, whether it is better
for the U.S. government to make such loans directly or instead to guarantee loans made by private
lenders depends upon whether its advantage in monitoring is partial or complete. Most likely,
however, this advantage is partial. While the government possesses some relevant information that
private lenders lack, private lenders may conversely possess some information that the government
lacks. For instance, such lenders may have greater expertise in project evaluation and risk
assessment, especially if the loan is intended (as in the case of the Mexican but not the Israeli loan
guaranties) for the purposes of economic development or financial reform. The participation of
private lenders at the underwriting stage, moreover, also helps to guard against the danger that the
government will make an economically unsound loan for unrelated (and possibly unwarranted)
political purposes. For all these reasons, then, the total costs of such a transaction are likely to be
minimized by structuring it as a guaranty, as opposed to an direct government loan.
B. The scope of defenses to liability under standby letters of credit

In recent years, an increasing proportion of commercial guaranties have been created through the
device of the standby letter of credit (SLC).115 The popularity of SLC’s is mainly due to two related
factors. First, the administrative costs of processing payment under a letter of credit are relatively
low. Under the law and traditional practice in the letter–of–credit field, a beneficiary C’s rights
against an issuer G depend only on its formal compliance with the conditions of the letter under which
it seeks payment. Ordinarily, these conditions are confined to the presentation of documents attesting
that C is entitled to draw on the credit; for this reason, letters of credit are also referred to as
“documentary credits.”
For example, under a traditional commercial letter of credit (CLC) designed to support an
international sale of goods, the buyer–applicant might engage an issuing bank to promise to pay the
seller–beneficiary upon presentation of documents showing that the goods have been shipped. The
required documents might typically include invoices, packing lists, bills of lading, etc.116 Under this
arrangement, all three parties are reasonably well protected against default. The seller can safely part
with the goods in advance of payment, because the issuer’s obligation to pay is absolute once the
documents have been properly presented. The buyer, conversely, is protected by the issuer’s
obligation to pay only against the requisite documents. And the issuer, for its part, is protected by
its control over the documents, which the buyer will need to retrieve the goods from the carrier, as
well as by the fact that the buyer often puts down cash up front to buy the letter of credit initially.
(Indeed, if the buyer does not put down cash, it will have to meet the issuer’s usual standards for a
short–term loan, complete with covenants and collateral).117 A standby letter of credit differs from
a traditional CLC only in the type of documents that must be presented at drawing; under a standby
letter of credit, the issuing bank promises to pay the creditor–beneficiary upon receiving documents
115. The dollar value of standby letters of credit issued by US and foreign banks increased from $51 billion in
1980 to $225 billion in 1986; and estimates of the current outstanding volume range between $400 and $500
billion worldwide and approximately $250 billion in the United States alone. See Boris Kozolchyk, The
Financial Standby: A Summary Description of Practice and Related Legal Problems, 28 U.C.C.L.J. 327, 328
(1996).
116. See generally James J. White and Robert Summers, UNIFORM COMMERCIAL CODE (4th practitioners’ ed.
1995, vol. 4), §26.1; John F. Dolan, FUNDAMENTALS OF COMMERCIAL ACTIVITY: A LAWYER'S GUIDE
(Boston: Little, Brown, 1991), ch. 5.
117. Of course, the buyer (or the issuing bank, in the unlikely event that the buyer is uncollectible) does bear the
risk that the documents, while conforming to the formal terms of the LC, inaccurately describe the goods that
have been shipped or that the goods themselves are defective or fail to conform to the underlying contract.
This risk is somewhat lessened by the (admittedly costly) option of suing the seller for breach of contract in its
home jurisdiction, however, and can be further minimized by proper specification of the requisite documents.
averring that the issuer’s standby obligation has come due. As with an ordinary CLC, the specific
documents to be presented depend on the terms of the letter, and could range from a final judgment
against the debtor as issued by a court of competent jurisdiction to the creditor’s sworn statement that
the debtor has defaulted on his underlying obligation to a mere sight draft.
Standby letters of credit, of course, support quite different types of economic transactions from
traditional CLC’s and thus contain different terms and payment conditions. For instance, SLC’s
typically involve higher fees, cover larger loans, have fewer restrictions on transfer, and provide for
a much longer expiration period than CLC’s.118 For the same reasons, banks issuing SLC’s usually
require different types of security from their applicants than they do when issuing traditional CLC’s.
Most obviously, it is not feasible to require a SLC applicant to pay cash up front, since the primary
point of the transaction is to provide him with liquid credit. Nonetheless, despite these economic
differences, the SLC offers a way for banks to adapt and apply their traditional expertise in
guaranteeing long–distance payment to a much wider set of business transactions, including real
estate developments, performance bonds, sales of corporate commercial paper, leases, private
placement of securities, and public offerings of municipal debt and other securitized assets.119 The
result is that SLC’s today dominate CLC’s in commercial importance.120
The characteristic feature that unites the standby and commercial letters of credit, however, and
the key to its low administrative costs, is that the conditions precedent for payment are essentially
documentary. As a result, C need not go to the expense of establishing the actual fact of D’s default
in order to collect from G under a letter of credit; and G need not investigate the substance of the
118. See Kozolchyk, supra, note 115, at 329, 334 (reporting author’s survey of practices at six major national
banks).
119. See John F. Dolan, THE LAW OF LETTERS OF CREDIT : COMMERCIAL AND STANDBY CREDITS, Boston:
Warren, Gorham & Lamont (2d ed 1991) §1.06 (also surveying uses of standby credits to support, inter alia,
tax shelters, payment obligations of foreign sovereigns, class–action settlements, interest rate swaps,
compliance with arbitral proceedings, and the child visitation rights of a divorced parent residing in a foreign
jurisdiction).
120. See Kozolchyk, supra, note 115, at 337 (estimating that quarterly average of new SLC issuance in 1994 was
approximately $140 billion, as opposed to $32 billion in total CLC’s outstanding). The ratio of SLC to CLC
activity is even greater than this estimate indicates, since SLC’s are typically outstanding for a longer period of
time than CLC’s.
underlying transaction before paying. Instead, C simply obtains the requisite documents and G checks
them against the formal requirements of the letter.121
The second major advantage of the SLC is that it allows the role of guarantor to be played by a
commercial bank. As a general matter, US banks are prohibited by a variety of federal and state
regulations from issuing ordinary contractual guaranties; and under most traditional interpretations
of bank charters, furthermore, issuing a guaranty would be regarded as an ultra vires (and hence
voidable) act.122 Such restrictions have traditionally been justified on the grounds that guaranties do
not accord with sound banking practice, cannot adequately be supervised under standard accounting
procedures, and pose too great a risk to solvency.123 They have also been motivated by independent
political goals, such as the populist desire to limit banks’ economic power and to prevent them from
competing with other financial institutions such as sureties and insurance companies.124
These general restrictions on bank guaranties, however, have not been interpreted to prevent
banks from offering standby letters of credit that are the functional equivalent. In large part this
distinction is an accident of history and of the deliberate exploitation of regulatory loopholes.
Commercial banks’ authority to issue letters of credit is well–established and uncontroversial, dating
back to the early days of the law merchant.125 Indeed, the traditional purpose of a letter of credit —
ensuring payment between merchants dealing at a distance — lies at the core function of banking.
While standby letters are used in different sorts of transactions from traditional CLC’s and present
different sorts of risks, they follow take the identical form as CLC’s and are often indistinguishable
from them on their face. This is especially so for “clean” letters of credit — that is, those that are
121. This central aspect of the letter of credit is part of its legal definition. UCC §5–102(a)(10), defining “letter of
credit,” and (a)(6), defining “document,” provide that letters of credit can condition payment only on the
presentation of specified records. Similarly, the Comptroller of the Currency’s Interpretive Ruling 7.7016, 12
CFR §7.7016 (1994), directs that “As a matter of sound banking practice, . . . the bank must not be called
upon to determine questions of fact or law at issue between the account party and the beneficiary.” But see
Kozolchyk, supra, note 115, at 358 (while criticizing such practice as legally unsound, indicating that
nondocumentary conditions in letters of credit are common).
122. See generally Richard A. Lord, The No–Guaranty Rule and the Standby Letter of Credit Controversy, 96
BANKING L.J. 46 (1979).
123. Id. at 56–57.
124. Id. at 56; see also Bank of Michigan v. Niles, 1 Doug. 401 (Mich. 1844).
125. See Rufus J. Trimble, The Law Merchant and the Letter of Credit, 61 HARV. L. REV. 981, 981–993 (1948).
payable upon the beneficiary’s mere statement that payment is due and presentation of a draft. The
law of letters of credit, accordingly, has drawn no distinction between CLC’s and SLC’s.126
As a practical matter, moreover, because letters of credit constitute a mercantile specialty, their
legal and economic consequences have historically been appreciated by only a small subgroup of
commercially sophisticated practitioners and judges. The esoteric quality of the standby letter
undoubtedly made it harder for generalist judges and regulators (not to mention the banks’ economic
competitors) from recognizing its potential as a device for offering guaranties. Instead, by the time
the commercial uses of the SLC became fully apparent in the late 1970's, the political climate for
banking regulation had drastically changed. The 1980's ushered in a period of general financial
deregulation, in which a host of traditional restrictions on commercial banks, ranging from limits on
branch banking to the prohibition of paying interest on demand deposits to restrictions on bank’s
entry into the credit card business, were loosened.127 As a result, whatever regulatory concerns may
have justified traditional restrictions on bank guaranties are today substantially tempered. With the
development and deployment of accounting techniques that better enable regulators and outside
investors to evaluate the risk of guaranties, furthermore, such restrictions are likely to become
increasingly obsolete with the passage of time.128
In light of these economic and historical facts, however, one can understand why the letter of
credit’s documentary nature, and its formal separation from the underlying transaction for which it
serves as security, are viewed by most specialists in the field as critical to its commercial utility. This
separation is referred to in the relevant literature and statutory commentary — and celebrated — as
the “independence principle.”129 On this orthodox view, keeping the issuing bank’s obligation to the
126. In this regard, UCC Article 5 does not even make reference to the term “standby letter of credit,” and does
not prescribe what specific documents might serve as a condition for drawing on a letter of credit.
127. See generally Daniel Fischel, Andrew M. Rosenfield, and Robert S. Stillman, The Regulation of Banks and
Bank Holding Companies, 73 VA. L. REV. 301 (1987).
128. See, e.g., the developments discussed in notes 50, 56, and 97, supra (describing accounting innovations
implemented under auspices of the Federal Credit Reform Act, Federal Reserve, Comptroller, Basle
Committee on Banking Regulation and Supervisory Practice, and FASB). In addition, allowing a bank to
increase its leverage by issing guaranties may help to counteract other incentive problems, such as the bank’s
incentive to pass up investments that, by reducing the risk of outstanding debt claims, effect a wealth transfer
from stockholders to debtholders. On this last possibility, see Christopher James, The Use of Loan Sales and
Standby Letters of Credit by Commercial Banks, 22 J. MONETARY ECON. 395 (1988).
129. See, e.g., Harfield, supra, note 13; Trimble, supra note 125, Alces, supra note 24.
beneficiary formally independent from the underlying transaction, and from any claims the issuer or
the applicant might have against the beneficiary arising out of that transaction, is essential to keeping
the costs of a letter of credit low and its market value high.130
The specialists’ concern for independence stems from considerations that are familiar from the
law of negotiable instruments and its doctrine of the holder in due course. If it were a defense to
issuer liability that the beneficiary had breached some substantive duty toward the applicant such as
shipping nonconforming goods or failing to provide promised security for its own performance, it
would no longer be possible to judge letters of credit on their face. Instead, a letter of credit’s value
would have to be discounted by the possibility that a substantive dispute would develop (or be
opportunistically manufactured) before drawing. This risk would not only make processing and
paying letters of credit more expensive; it would also substantially reduce the beneficiary’s certainty
of receiving payment. For this reason, while modern letter of credit law does not entirely exclude
substantive considerations arising out of the underlying contract, it tries to keep them to a minimum.
Under the 1995 revision of Article 5, for instance, while an issuer retains the discretion to refuse to
pay claims that are materially fraudulent or based on forged documents, and while applicant may be
entitled to enjoin payment on proof of such fraud or forgery, the standards for establishing such a
defense or obtaining such an injunction have been substantially heightened.131
For our purposes here, however, the main relevance of the independence principle is that it has
served to wall off the doctrinal and policy analysis of letters of credit from that of guaranties.
Throughout their writings, letter–of–credit specialists repeatedly stress that while the standby letter
might serve some of the economic functions of a guaranty, it is not a guaranty, it is not a type of
guaranty, it is a grave analytic error to call it a guaranty, and most importantly, it is not subject to the
law of suretyship.132 The specialists’ insistence on this distinction, however, is hardly a matter of
130. The drafters of revised Article 5 took pains to make this point with special clarity. As that article’s Prefatory
Note states: “Revised Article 5 clearly and forcefully states the independence of letter of credit obligations
from the underlying transactions that was unexpressed in, but was a fundamental predicate for, the original
Article 5. . . Certainty of payment, independent of other claims, setoffs or other causes of action, is a core
element of the commercial utility of letters of credit.”
131. Compare revised §5–109 (requiring “material fraud by the beneficiary on the issuer or applicant”) with old
§5–114 (employing broader and vaguer standard of “fraud in the transaction”).
132. See, e.g., White and Summers, supra, note 116, at 117 (“In short, a letter of credit is a letter of credit. As
Bishop Butler once said, ‘Everything is what it is and not another thing.’ ”)
abstract formalism; nor is it solely motivated by the aforementioned regulatory loopholes that allow
commercial banks to issue letters of credit but not guaranties. Rather, the specialists also wish to
protect the certainty and reliability of letter of credit transactions. Walling off standby letters of credit
from suretyships achieves this result in two ways.
First, as we have seen, suretyship law typically allows guarantors to raise a broad variety of
substantive defenses to liability, including both the claims of the primary obligor, and the special
suretyship defenses that are based on actions of the creditor that may have increased the risk of
default. While it is possible to contract out of these defenses — for example, by having the guarantor
waive them in advance — such arrangements are often disfavored by courts and regulators. Even
when such waivers are enforceable, furthermore, having to arrange for them adds to the cost of the
overall transaction.
Second, suretyship law, as a branch of the common law, includes among its doctrines a variety
of general equitable and quasi–equitable principles such as reasonableness, due care, waiver, estoppel,
and the like. These doctrines make suretyship law much less rule–oriented than letter of credit law,
much more subject to judicial expansion in the face of sympathetic facts, and hence much less
predictable to commercial specialists. Such specialists and their clients are repeat players in the
guaranty game; and as such they place high value on knowing what their rights and duties are and
avoiding interpretative disputes. From their perspective, applying general common–law principles
in the letter of credit field merely threatens to disrupt a clear and well–functioning set of specialized
usages and commercial practices.
For these reasons, representatives of issuer and beneficiary interests fought hard — and for the
most part successfully — to keep letters of credit outside the scope of the new Restatement of
Suretyship and Guaranty. As a result of their efforts, §4(2) of the Restatement explicitly states that
obligations governed by the law of letters of credit are excluded from its coverage.133 To the same
133. See Restatement (Third) of Suretyship and Guaranty, §4, comment c (“The reason for this exclusion is
practical. The law governing letters of credit is quite well developed, and is generally understood to govern all
letters of credit, both traditional and standby. No good purpose would be served by disturbing that state of
affairs. To the extent that the law governing letters of credit is silent as to a particular issue, however,
suretyship law is a potential source of generally appropriate analogies.”) For an insider’s account of the
politics of drafting the Restatement, see Rapson, supra, note 1; for an analogous account of the Article 5
revision process, see James J. White, The Impact of Internationalization of Transnational Commercial Law:
The Influence of International Practice on the Revision of Article 5 of the UCC, 6 J. INTERNATIONAL. L. BUS.
189(1995).
end, the new 1995 revision of Article 5 goes out of its way to stress that letters of credit constitute
a distinct commercial specialty not governed by ordinary principles of contract or suretyship;134 and
also introduces several changes from prior law that are plainly intended to further formalize the
parties’ rights and duties and to discourage interpretative disputes.135
Other legal commentators in the guaranty field, however, especially some of those involved in the
preparation of the new Restatement, maintain that it was a mistake to treat standby letters of credit
and guaranties as conceptually distinct transactions subject to wholly separate theories of liability.
Peter Alces, for instance, has argued that the difference between standby letters of credit and
guaranties is merely a matter of degree and that, accordingly, the compromise that excluded SLC’s
from the scope of the new Restatement was unprincipled and unjustified.136 As Professor Alces points
out, because SLC’s remain subject to the defense of material fraud, it is misleading to view them as
entirely independent of the substantive transactions that underlie them. Even though fraud may be
difficult to prove, determining whether it has occurred necessarily requires reference to the substance
of the underlying transaction; and materiality is always in the eye of the beholder. Conversely, the
parties to a letter of credit are always free to specify documentary conditions that turn on underlying
substance, such as certificates evidencing independent third parties’ determinations of fact or law.
For these reasons, he argues, fraud and simple breach of contract — and thus independence from the
underlying transaction and interdependence with it — are most usefully understood as points along
a single continuum.
134. As comment 1 to revised §5–103 states: “This article does not apply to secondary or accessory guarantees and
it is important to recognize the distinction between letters of credit and those guarantees. It is often a defense
to a secondary or accessory guarantor's liability that the underlying debt has been discharged or that the debtor
has other defenses to the underlying liability. In letter of credit law, on the other hand, the independence
principle recognized throughout Article 5 states that the issuer's liability is independent of the underlying
obligation. . . Only staunch recognition of this principle by the issuers and the courts will give letters of credit
the continuing vitality that arises from the certainty and speed of payment under letters of credit. To that end,
it is important that the law not carry into letter of credit transactions rules that properly apply only to
secondary guarantees or to other forms of engagement.”
135. These changes include defining good faith as mere “honesty in fact,” in contrast to the more equitable
standard of “reasonable commercial standards of fair dealing” found in Article 2 and the proposed revision of
Article 1 [see new §5–102(a)(7)]; substituting a strict compliance principle for the doctrines of waiver and
estoppel that would otherwise apply under section 1–103 [new §5-108(a)]; eliminating consequential damages
and, more strikingly, the duty to mitigate damages [new §5-111(a)]; and providing that courts must award
attorneys’ fees to the prevailing party in any dispute that arises over the issuer’s duty to pay [new §5-111(e)].
136. See Alces, supra, note 24.

To Professor Alces’ arguments we may add the observation that even if the status of the parties’
underlying transaction is irrelevant to the issuer’s immediate duty to honor a standby letter of credit,
it is still quite relevant to whether the beneficiary is ultimately entitled to retain payments r eceived
under that letter. Even under revised Article 5, a beneficiary who draws on a letter of credit impliedly
warrants (to the issuer) that no material fraud or forgery has taken place and (to the applicant) that
the drawing does not violate the underlying agreement between beneficiary and applicant.137 For
instance, if the underlying contract authorized the beneficiary to draw on the letter of credit only upon
the applicant’s default, a beneficiary who drew in the absence of default (or in the mistaken belief that
there had been default) would be obligated to return to the applicant any funds wrongfully obtained
and to make him whole for any additional losses that had forseeably resulted from the wrongful draw.
Similarly, if a standby letter of credit required the beneficiary to present a sworn affidavit that it had
exercised best efforts to collect from the applicant or levy on his collateral, and it knowingly and
falsely so swore, the issuer would be entitled to the return of any funds advanced on the basis of the
false affidavit.
Thus, while the letter of credit places the burden of inertia on the applicant in the event of a
substantive dispute, and while it may force the applicant and issuer to pursue the beneficiary in a
distant and inconvenient jurisdiction, it does not insulate the beneficiary from its obligations to comply
with its substantive contractual duties. To restate Alces’ point in somewhat stronger terms, the
difference between standby letters of credit and guaranties is not substantive, but procedural. From
a functional vantage point, and from the viewpoint of parties choosing what terms to include in their
contractual arrangements, guaranties and SLC’s are analogous devices. Indeed, the drafters of
revised Article 5 conceded the analogy with in their treatment of the specific question of whether an
issuer who extends funds should be subrogated to the beneficiary’s rights against the applicant. Prior
case law, while not unanimous on the point, had tended to deny subrogation on the grounds that
denial was necessary to maintain the formal distinction between letters of credit and guaranties.138
Revised section §5–117, however, provides that an issuer who honors a beneficiary’s presentation,
137. See revised §5-110(a); compare previous §5-111.
138. See generally Amelia H. Boss, Suretyship and Letters of Credit: Subrogation Revisited, 34 WM. & MARY L.
REV. 1087 (1993) (criticizing the case law and defending the approach of revised Article 5).
and an applicant who reimburses his issuer, will be subrogated to the same extent they would be if
the transaction were a suretyship.139
With all this as background, we are now in a position to ask: what bearing does our general
economic framework, as laid out in the previous sections of this paper, have on this debate? What
implications does it suggest for the possible conditions that might be put on the liability of
letter–of–credit issuers?
First of all, it suggests that Professor Alces is correct that the independence principle so sacred
to letter–of–credit specialists is ultimately just a shibboleth. While it might well be appropriate to
limit issuers’ defenses to liability — just as it is sometimes appropriate to limit or waive guarantors’
defenses under suretyship law — any difference in legal treatment between standby letters of credit
and conventional guaranties ought to turn on functional differences in the nature of the risks being
allocated and on the particular monitoring and liquidity obligations being assigned to the various
participants, not on abstract formalisms. Furthermore, while there do exist some functional
differences between guaranties and standby letters of credit, those differences are smaller than the
differences between standbys and traditional CLC’s.
More specifically, the configuration of monitoring and liquidity costs under a standby letter of
credit is closer to that of a conventional guaranty than to that of a traditional letter of credit. This
is true for two main reasons. First, letter–of–credit specialists typically regard the administrative
costs of processing payment as the primary factor determining a letter’s commercial value.140 This
view makes good sense in the case of a traditional commercial letter of credit, where payment is
intended to occur in the ordinary course. With a standby letter of credit, however, the parties’
ordinary expectation is that the applicant will make good on his underlying obligation and the letter
will never need to be drawn against. While the specialists acknowledge this distinction (and, often,
draw attention to it), most of them tend to fail to recognize its main economic implication — that
costs of payment make up a much smaller fraction of the issuer’s expected transaction costs under
a standby letter of credit, and are thus far less important to the standby letter’s commercial value.
139. See §5–117, official comment 1 (specifically referencing the Restatement (Third) of Suretyship and
Guaranty).
140. See, e.g., White and Summers, supra, note 116, at __; Dolan, supra, note 116, at __; Kozolchyk, supra, note
115, at __.
For standbys, these payment costs are dwarfed by the costs of investigation, underwriting,
monitoring, collection, and the bearing of losses following default.
Second and similarly, in the event that a standby letter does come to be drawn on, there is a much
higher chance than under a traditional CLC that the applicant will be uncollectible.141 Because of the
substantive nature of the transactions supported by standby letters, standbys are typically outstanding
for a much longer period of time than CLC’s and do not require the applicant to put down cash up
front. As a result, the beneficiary’s ability and incentives to monitor the applicant are much more
important in the standby than in the traditional context.
Admittedly, with some types of standby letters, the beneficiary’s ability to monitor is very limited,
as in asset securitization arrangements in which a bank guaranty effectuated through a standby letter
of credit allows mortgage– or municipal–bond–backed securities to be marketed to small and distant
investors.142 For other types of SLC’s, however, the beneficiary may well be in a position to do some
part of the monitoring. The fact that a bank has a comparative advantage in assessing its customer’s
overall creditworthiness and collecting from him in the event of default, for instance, does not
necessarily imply that it is the best underwriter with regard to the line of business in which he now
proposes to invest. Assigning at least some monitoring tasks to the beneficiary, accordingly, and
backing up that assignment with some legal consequences if those tasks are mishandled, may well
lower the total expected costs of the underlying loan.
The default rules and understandings of traditional letter of credit law, however, are motivated
by the needs of the traditional CLC transaction, and as such do not provide beneficiaries of SLC’s
with particularly good incentives to engage in monitoring. Under a traditional CLC, the main task
of the beneficiary is to ship conforming goods, not to monitor the applicant’s ability to pay. The
beneficiary’s main incentives to perform its assigned task, furthermore, do not come from any promise
or warranty it makes to the issuer, as material fraud in the transaction is difficult to prove and the
issuer will likely be looking primarily to the applicant for reimbursement. Rather; its incentives come
from the fact that if the beneficiary does not perform it will be liable to the applicant for breach of the
underlying contract, subjecting it to damages in its home forum. A breaching beneficiary,
141. See White and Summers, supra note 116, at __; Dolan, supra, note 116, at __.
142. See generally Kozolchyk, supra, note 115, at 338–354 (describing typical municipal bond transaction in
detail).
furthermore, is also likely to forfeit future business from the applicant and possibly from his business
acquaintances.
Under a SLC, in contrast, these enforcement devices are substantially weakened. The insolvency
of the applicant reduces the chance that it will pursue its legal rights against the beneficiary in a
distant forum, and largely eliminates the factor of future business. More importantly, the beneficiary’s
failure to engage in efficient monitoring will probably not even be a breach, for its main legal duty to
the applicant is to provide him with liquid funds, not to monitor him.143 Indeed, the benefit of such
monitoring accrues primarily to the issuer rather than the applicant. But under a standard letter of
credit, standby or otherwise, the beneficiary owes the issuer no legal duty to monitor. At most, the
beneficiary is obligated not to engage in outright fraud and not to present materially false or forged
documents when drawing on the letter.
Based on the foregoing analysis, therefore, I tentatively conclude that Article 5's default
provisions regarding liability, defenses, and risk allocation are not optimally suited to the stereotypical
SLC transaction. Such a conclusion should not be regarded as especially surprising, given that the
default rules of letter of credit law were developed and refined long before SLC’s were invented, and
that the transactions governed by SLC’s have been cast in that form in large part to evade regulatory
restrictions that, while possibly justifiable in their original historical context, are in today’s financial
markets increasingly obsolete. If SLC were today being designed from scratch, they might well be
structured as ordinary guaranties and governed by the Restatement. To the extent that they include
significant nondocumentary conditions precedent to payment, in fact, they are already governed by
the Restatement notwithstanding the intentions of their drafters.144
This conclusion, however, does not necessarily imply, pace Professor Alces, that Article 5 needs
to be amended or that all standby letters of credit should have been made subject to the new
143. But see Daniel R. Fischel, The Economics of Lender Liability, 99 YALE L.J. 131, __ (1989) (surveying case
law under which inept debtors were allowed to recover from creditors who should have known better than to
lend to them.)
144. See revised §5-102, comment 6, which states that: “when a document labelled a letter of credit requires the
issuer to pay not upon the presentation of documents, but upon the determination of an extrinsic fact such as
applicant's failure to perform a construction contract, and where that condition appears on its face to be
fundamental and would, if ignored, leave no obligation to the issuer under the document labelled letter of
credit, the issuer's undertaking is not a letter of credit. It is probably some form of suretyship or other
contractual arrangement and may be enforceable as such.”
Restatement. Many of Article 5's rules are default rules, after all, subject to change by the contracting
parties if they choose. The process for amending the UCC is cumbersome, and as a practical matter
requires the substantial assent of the main interest groups affected.145 Furthermore, the preference
of letter–of–credit specialists for rules over standards and for well–established trade usages over
contextual common–law principles is a legitimate one.
The conclusion does suggest, however, that parties transacting under Article 5 should give serious
thought to developing contractual provisions and documentary conditions that more closely resemble
the substantive risk allocations and monitoring duties provided under the Restatement for
stereotypical guaranties. For instance, parties drafting loan contracts intended to be supported by a
standby letter of credit might consider including a term stating that the issuer of the standby letter is
to be considered a third–party beneficiary of any promises made by the lender/beneficiary to the
borrower/applicant. This clause would allow the issuer to assert claims for breach of such promises
under the underlying contract, even though it could not independently assert any such claim under
the letter of credit. Similarly, a bank drafting a standby letter of credit might consider including
among its documentary conditions the requirement that the beneficiary present a certified judgment
of an independently selected arbitrator that the applicant is actually in default, or alternatively, a
sworn affidavit, signed by the beneficiary, stating that it has undertaken all commercially reasonable
efforts to monitor the applicant and maintain all rights against collateral during the period leading up
to default. Similar affidavits dealing with underwriting and credit investigation could also be required
at the outset of the transaction as a condition for issuing the standby letter in the first place.
Parties who attempt to draft such arrangements, of course, may well ultimately decide that the
arrangements are not worth bothering with, or may discover that beneficiaries are unwilling to agree
to them. But such an outcome should only follow if the costs such arrangements impose on
beneficiaries exceed the savings they create for applicants and issuers. If the beneficiaries of standby
letters of credit do have a comparative advantage in some aspects of monitoring, on the other hand,
they should be willing to accept such provisions in exchange for changes in other terms of the bargain
that benefit them — such as an increase in the rate of interest they receive for lending funds.
Similarly, such provisions and conditions should not be inherently objectionable to bank
regulators, since if the participants were to agree to them they would operate to reduce issuers’ risk
145. See generally Alan Schwartz and Robert E. Scott, The Political Economy of Private Legislatures, 143 U. PA.
L. REV. 595 (1995)
rather than increase it. In this regard, that banks originally began writing SLC’s in order to evade
prohibitions on their issuance of guaranties constitutes a significant irony. To the extent that the
beneficiaries of standby letters of credit owe issuers duties of care, and to the extent that those issuers
can enforce duties owed by beneficiaries to applicants, bank solvency will only be strengthened. By
the same token, bank depositors and the FDIC, though perhaps not the banks’ business competitors,
will only benefit.
V. Conclusions
This paper has demonstrated that the stereotypical guaranty transaction is motivated by a basic
economic logic; its thesis is that a short and simple list of principles can explain the structure and
purpose of such transactions across a wide variety of legal and policy contexts. To recapitulate the
main argument, supplying credit entails various costs: including monitoring costs (defined here to
include such things as investigation, risk–assessment and underwriting, auditing, insurance, collection
and salvage), liquidity costs, miscellaneous administrative costs, and the time value of money. In an
ordinary direct loan, the creditor incurs all these costs and charges the debtor for them in the form
of interest. A guaranty, in contrast, provides a way for a lender to unbundle some of the risks of the
debtor’s default, together with associated monitoring obligations, and sell it to another party. This
other party is the guarantor. Such an arrangement makes sense if and only if the guarantor has a
comparative advantage in monitoring, and the original lender has a comparative advantage in
liquidity; else it will be cheaper and more efficient for either one just to lend on an unconditional basis.
In cases where these conditions are met, however — or more precisely, where the advantages of
specialization are large enough to warrant incurring the additional transaction costs of writing and
enforcing a three–corner arrangement — such unbundling can lower the total costs of making the
loan, creating a transactional surplus that can be shared among all three participants.
The paper also explained how and when guaranties are superior, as measured by the criterion of
economic efficiency, to alternative three–corner transactions such as straight intermediation. The
distinctive feature of a guaranty relative to intermediation is that the creditor buys an option to collect
directly from the debtor, instead of collecting from the guarantor and having her collect from the
debtor. Granting the creditor such an option is efficient when, and only when, the creditor holds a
comparative advantage in some aspect of monitoring, relative to the guarantor or to her other
creditors. Otherwise, the rights that attach to the option will be more valuable remaining in the
guarantor’s hands.
The foregoing analytic framework helps us to integrate and synthesize a disparate body of
knowledge from multiple fields of legal and commercial practice — including, inter alia, personal
property security, mortgages, negotiable instruments, corporations, letters of credit, the law of
contract assignment and delegation, government contracts, and Federal expenditure programs aimed
at encouraging the greater availability of credit. In so doing, it also enables insights from one field
to be translated and analogized for the purposes of critiquing and improving transactional planning
in others. In the previous section of the paper, for instance, we saw that our framework implies that
government loans and guaranties are inefficient unless the government has some comparative
advantage in monitoring debtors, and that the default rules provided by U.C.C. Article 5 probably do
not provide an optimal mix of rights and duties in the context of standby letters of credit.
In addition, though we did not focus on the possibility here, the framework could also be used
to help clarify ambiguous transactions for the purposes of other social policies or legal doctrines. As
just one illustration, courts applying the common-law Statute of Frauds have often found it difficult
to tell whether a given transaction should be deemed a suretyship, and hence subject to the formal
requirement of a signed writing, or rather an agreement to engage in financial intermediation,
enforceable even when oral.146 The analysis above, however, suggests an answer to this problem. As
we have seen, the functional difference between a guaranty and straight intermediation is that under
the former, the creditor C has some comparative advantage in monitoring the debtor D, and thus
retains the obligation and incentive, legal or practical, to engage in such monitoring. If C cannot
plausibly be assigned any type of monitoring obligation, conversely, the transaction is not properly
a guaranty from the economic viewpoint, notwithstanding that the parties may have denominated it
as such or formally preserved C’s right to proceed against D. This economic distinction, furthermore,
accords with the functional purposes of requiring guaranties but not intermediation agreements to be
put into writing. An intermediary G who knows that she is the only person with the capacity or
incentive to engage in monitoring the debtor is likely to take the prospect of responsibility for his
default very seriously. If G assumes that the ultimate creditor C will undertake a significant portion
of the risk and the associated monitoring, however, she may enter into the transaction with much less
forethought. Requiring a writing in this latter case, accordingly, promotes the overall cautionary and
channeling purposes of the Statute.147
Other extensions and potential applications of the framework abound, for the basic principles of
the guaranty contract touch virtually every field of commerce and industry. The factors that influence
whether a guaranty is efficient, for instance, are analogous to those that arise when analyzing
problems of reinsurance, financial options, trusts, and the practice of asset securitization. Similarly,
146. See generally Charles C. Marvel, Annotation, Promise by One Other than Principal to Indemnify One
Agreeing to Become Surety or Guarantor as Within Statute Of Frauds, 13 A.L.R.4th 1153 (1996). Thanks to
Allan Axelrod for suggesting this application.
147. Cf. Arthur Von Mehren, Civil–Law Analogues to Consideration: An Exercise in Comparative Analysis, 72
HARV. L. REV. 1009, 1016 et seq. (1959) [discussing evidentiary, cautionary, channeling and deterrent
functions of the common–law consideration doctrine].
the incentive properties of suretyship are akin to those provided by the doctrines of fraudulent
conveyance law; indeed, with only a small stretch the duties imposed on transferees by such doctrines
could be reconceived as those of a guarantor. And as the discussion in subsection III.D, supra, made
plain, the economic incentives provided under guaranty relationships also closely resemble those of
secured credit. In this last regard, it is additionally noteworthy that guaranties and security interests
are commonly used in tandem. It would be worth exploring in subsequent research the close
connection between these alternate devices, and the extent to which they may operate as substitutes
or complements in particular instances.
Furthermore, this paper has focused almost entirely on the problem of minimizing transaction
costs for the three parties to the potential guaranty. As the introductory discussion to section II
above indicated, however, guaranties also can impose significant external costs on competing
creditors and debtors. Now that the basic economics of the guaranty transaction have been laid out,
an obvious next step would be to consider how such third–party externalities could best be addressed
and internalized. Possibilities in this regard might include improved accounting standards, stronger
financial disclosure requirements, or a centralized filing system for guaranties on the model of the one
currently provided for security interests under Article 9 A full discussion of such alternatives,
however, and of how they might play out in the multiplicity of contexts in which guaranties are used,
is better saved for another paper and another day.
In short, the importance of the guaranty contract in law and commerce is vast — so vast, in fact,
that it is remarkable that its incentive structure has apparently never been investigated using the tools
of the economic analysis of law. The actual amount of economics needed for the task, furthermore,
is not especially complicated. Once one asks the basic question — why a guaranty? — and draws
the connections between the various contexts in which the guaranty relationship arises, the rest of the
analysis follows in fairly straightforward fashion. The primary contribution of this paper, accordingly,
is to have asked the question and drawn the connections.

An Economic Analysis of The

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

An Economic Analysis of The

Uploaded by

Copyright:

Available Formats

The Center for Law and Economic Studies

Columbia University School of Law

An Economic Analysis of the Guaranty Contract

Working Paper No. 136

A revised version of this working paper has been published in

Please send comments to:

Do not quote or cite without author's permission.

This paper can be downloaded without charge at:

The Columbia Law School Working Paper Series Index:

The Social Science Research Network Electronic Paper Collection:

Avery Wiener Katz

Draft version: March 1998

Keywords: Contracts, commercial law, guaranties and suretyships, risk allocation.

II. The law of guaranties and suretyships: a sense of the terrain

A. The stereotypical guaranty relationship

Figure 1: Terminology of the guaranty relationship in alternative contexts

3. Standby letter of credit

5. Statutory and fidelity bonds

6. Recourse assignment of accounts, chattel paper, or instruments

7. Assumption of leases and real estate mortgages

8. Joint and several liability

A number of U.S. government–chartered and sponsored institutions participate directly in credit

17. Id. at __.

10. Guaranties between independent national sovereigns

B. The law governing guaranties

26. See Restatement (Third) of Suretyship and Guaranty, §29 (1996).

27. See id. §21.

28. See id. §34.

29. See id. §36.

30. See id. §37 et seq. (“Impairment of Suretyship Status”).

31. Id. §47 (1996).

36. See UCC §3-419(d).

III. Guaranties and their alternatives: the choice of transactional form

53. Id., at __.

Figure 2. Alternative configurations of credit

A. Guaranties, unguaranteed lending, and intermediation

B. The cost of credit in a guaranty transaction

Similarly, controlling shareholders hold a comparative advantage in investigating and supervising

C. Why does the guarantor guarantee rather than simply lend?

71. See notes 55–57, supra, and accompanying text.

D. Why doesn’t the guarantor simply act as an intermediary?

88. See Schwartz, A Theory of Loan Priorities, supra, note 3.

IV. Applications, public and private

A. The government as guarantor

95. Id., at 507–509.

98. Id. §661(c).

99. Id. §661c(c)(1).

101. See discussions at notes 50, 51, and 56, supra.

111. Thanks to Scott Masten for suggesting this point.

B. The scope of defenses to liability under standby letters of credit

123. Id. at 56–57.

136. See Alces, supra, note 24.

137. See revised §5-110(a); compare previous §5-111.

You might also like